Skip to main content

Full text of "Meaning and change of meaning: with special reference to the English language"

See other formats


Digitized by tine Internet Arciiive 

in 2011 witii funding from 

LYRASIS IVIembers and Sloan Foundation 

and Change 
of Meaning 

Indiana University Studies 

in the 

History and Theory of Linguistics 

Editorial Committee: 

Dell H. Hymes, University of California, Berkeley 

John Lotz, Columbia University 

Thomas A. Sebeok, Indiana University 

Rulon Wells, Yale University 


and Change 
of Meaning 



First published 1931. Copyright Gustaf Stern 1931. 

Library of Congress catalog card number: 64-12351 
Manufactured in the United States of America 

My sincere thanks are due to the Curators of the two Founda- 
tions that have generously contributed to the cost of printing 
the present work, to the Committee of Goteborgs Hogskolas Arsskrift 
for including it in their series, and to Mr. G. Harward, B. A., who 
has kindly read a proof and revised my English. 

I wish also to acknowledge my great indebtedness to the New Eng- 
lish Dictionary, without the help of which a book of this ^ kind cannot 
be written. All quotations with a date but no reference are taken 
from the NED, except a few that are quoted from Matzner's Worter- 

The MS. of my book was completed about two years ago, and I have 
therefore only occasionally been able to take into consideration litera- 
ture published later than 1929. 

G. S. 


Ch. I. Introduction. 

The Position of Semasiology in Relation to Linguistics and Psychology i 

Methods of Work i 

Previous Work 4 

The Necessity for Systematic Work 7 

Plan of the Work 8 

The Linguistic Material 11 

The Psychological Theories. Final Remarks 14 

Ch. n. Preliminary Statements. 

2.1. Language and its Functions 17 

2.1 1. Speech and the Language System 17 

2.12. The Functions of Speech 18 

2.13. The Functions of Words and the Functions of Speech 21 

2.2. Signs and Meaning in General 23 

2.21. General Theory of Signs 23 

2.22. Signals, Symbols, and Substitute Signs 26 

Ch. m. The Definition of Verbal Meaning. 

Factors of the Meaning Complex 29 

Three Factors , 29 

The Word 31 

The Referent 31 

Word and Meaning 32 

Referent and Meaning 33 

Mental Content and Meaning 34 

The Definition of Meaning 35 

The Connection between Word and Meaning 35 

The Interrelation of the Three Factors 37 

The Objective Reference 38 

The Subjective Apprehension 41 

Tlie Traditional Range 42 

Summary. A Fourth Factor? 43 

Definition 45 


.1. i 

3. II. 







.2. 1 









Ch. IV. Analysis of Meaning. 

4.1. Psychic Elements and Categories of Meaning 46 

4.1 1. Psychological and Logical Points of View 46 

4.12. Cognitive and Emotive Elements 46 

4.13. Thought and Imagery 47 

4.14. Images 49 

4.141. Useful, Inadequate, and Irrelevant Images 49 

4.142. Images as Signs, Meaning, or Context 51 

4.15. Emotional and Volitional Elements of Meaning 54 

4. 151. Preliminary Remarks 54 

4.152. Methods of Expressing Emotive Elements 54 

4.153. Emotion and Communication 56 

4.154. Sources of the Emotive Elements: Permanent Elements 57 

4.155. Sources of the Emotive Elements: Incidental Elements 59 

4.156. Volitional Attitudes 60 

4.16. Central and Peripheral Elements of Meaning ^ 60 

4.17. The Vagueness of Meanings 63 

4.2. Logical Elements and Categories of Meaning 68 

4.21. Introductory 68 

4.22. Actual and Lexical Meaning. Isolated Words 68 

4.23. General and Particular Meaning 70 

4.24. Specialized and Referential Meaning 72 

4.25. Tied and Contingent Meaning 74 

4.26. Basic and Relational Meaning 75 

4.261. Introductory Remarks 75 

4.262. Psychic Reality of Relations. Relations as Referents 76 

4.263. Syntactical Relational Meaning 78 

4.264. Derivational Meaning 80 

4.265. Indicated Puport 81 

4.27. Word-meaning and Phrase-meaning 83 

4.28. Autosemantic and Synsemantic Meanings 85 

4.3. A Note on Other Definitions 87 

Ch. V. The Production of Speech. 

5.1. Introductory Remarks 90 










Research on Aphasia 90 

21. Preliminary Remarks 90 

22. Classification of Speech Defects 93 

23. Verbal Defects 94 

24. Syntactical Defects 95 

25. Nominal Defects 96 

26. Semantic Defects 98 

27. Remarks on Head's Theory 99 

Evidence of Normal Psychology 102 

31. Sentence Definitions. Paul and Wundt 102 


5.32. Sentence Formulation according to Selz 103 

5.33. Analytic Formulation 103 

5.34. Gradual Formulation 105 

5.35. Synthetic Formulation 108 

5.36. Remarks on Selz' Theories no 

5.37. Linguistic Automatism in 

5.4. Temporal Relations in Speech Production 114 

5.5. The Relation of Speech to Thought, and the Selection of Words in 

Speaking 120 

5.51. The Relation of Speech to Thought. Discreteness 120 

5.52. The Selection of Words in Speaking 123 

5.53. The Reaction of Words on Thought 126 

Ch. VI. The Comprehension of Speech. 

6. 1 . Visual and Auditory Perception of Words 129 

6.2. The Comprehension of Isolated Words 131 

6.21. The Act of Understanding in General 131 


Experimental Evidence 131 

Conclusions 137 

6.3. The Comprehension of Speech (Ordinary Discourse) 139 

6.31. Preliminary 139 

6.32. Context: Verbal, Perceptual, and Mental 139 

6.33. Active Interpretation 143 

6.34. Spearman's Analysis of Comprehension 147 

6.35. Biihler's Experiments on the Comprehension of Sentences 149 

6.36. Variations in the Process of Understanding 150 

6.37. Mistakes in Anticipation 154 

6.38. Word-meaning and Phrase-meaning 155 

6.4. Temporal Relations in the Process of Comprehension 157 

6.5. Intended and Comprehended Meaning. Identity and Discrepancy 159 

Ch. Vn. General Theory of Sense-change. 

7.1. Definition and Preliminary Delimitation 162 

7.1 1. Change and Stability 162 

7.12. Definition of Sense-change (Change of Meaning) 163 

7.13. Change and Fluctuation 163 

7.2. The Seven Classes of Sense-change 165 

7.21. The Fundamental Point of View 165 

Substitution 166 

Analogy 166 

Shortening 167 

Nomination 167 

Transfer 168 

Permutation 168 

Adequation 168 


Class I. 


Class II. 


Class III. 


Class IV. 


Class V. 


Class VI. 


Class VII. 


7.3. General Causes and Conditions of Sense-change 169 

7.31. The Three Immediate Causes of Sense-change 169 



Further Causes: External and Linguistic 169 

Further Causes: Intentional and Unintentional Processes 170 

The Ultimate Causes of Sense-change 171 

The General Conditions of Sense-change 174 

Scheme of Classification, and Final Remarks 175 

7.4. Some Relevant Problems 176 

The Spread of New Meanings 176 

The Transmission to New Generations 178 

The Part of Emotions in Sense-change 182 

Conservative Factors 184 

The Regularity of Sense-change. Semantic Laws 185 

Ch. Vm. Class I. Substitution. 

8.1. Theoretical Discussion 192 

8.2. Factual Change of the Referent 194 

8.3. Change in Knowledge of the Referent 197 

8.4. Change of Attitude to the Referent 198 

Ch. IX. Class n. Sense-changes due to Analogy. 

9.1. Group Formation 199 

9. 11. Preliminary Remarks 199 

9.12. Paul's Theory of Analogy 200 

9.13. Basic (Material) Groups 202 

9.14. Relational (Formal) Groups 204 

9.15. Conclusions 206 

/ 9.16. The Three Types of Analogy 207 

9.2. Combinative Analogy 207 

9.21. Esper's Experiments 207 

9.22. Application to Language 209 

9.23. A Question of Principle 212 

9.24. Flexional Groups 214 

9.25. Derivational Groups 216 

9.3. Correlative Analogy 218 

9.31. The Psychic Process 218 

9.32. Within one Language 219 

9.33. Two Languages involved (Sense-loans) 220 

9.331. Theoretical Discussion 220 

9.332. Single Words: Semantic Similarity only 221 

9.333. Single Words: Semantic and Phonetic Similarity 223 

9.334. Phrases 224 

9.34. The Problem of Synonyms 224 

9-35- Wundt's Korrelative Laut- und Bedeutungsanderungen 228 

9.4. Phonetic Associative Interference 230 

9.41. Preliminary Remarks 230 


g.42. Phonetic Resemblance as a Cause of Association 231 

9.43. Change of Meaning but not of Referent 233 

9.44. Change of both Meaning and Referent 233 

Cb. X. Class m. Sense-changes due to Shortening. 

10. 1. Theoretical Discussion 237 

10. 1 1. Preliminary Remarks. Habitual and Occasional Combinations. 

Individual and Typical Shortening. Clipping and Omission.... 237 

10.12. Criteria: The Referent 238 

10.13. Criteria: The Mental Content 239 

^i) The Logical Point of View 240 

(2) The Functional Point of View 241 

10.14. Criteria: The Sentence-Scheme 243 

10.15. Criteria: Historical (Individual) Shortening 244 

10.16. The Psychic Process 245 

10.17. The Conditions of Shortening 248 

10. i8. The Causes of Shortening 250 

10. 1 81. Preliminary Remarks 250 

10.182. Phonetic Causes 250 

10.183. Graphic Causes 253 

10.184. Functional Causes 256 

10.185. Economic Causes 257 

102. Classification of Shortenings 258 

10.21. The Principal Point of View. Clipping and Omission 258 

10.22. Permanent and Contextual- Sense-change. Delimitation of the 

Problem 260 

10.3. Clipping 261 

10.4. Omission 265 

10.41. Preliminary Remarks 265 

10.411. The Form of the Shortening. Contrasted Referents 265 

10.412. Omission of Qualifier and Genus pro Specie 267 

10.413. Further Delimitation 269 

10.42. Omission of Qualifier in Binary Combinations 271 

10.421. (Attribute) -|- Noun 271 

10.422. Noun -f (Prepositional Phrase) 272 

10.423. Verb + (Object or Complement) 272 

10.424. Adjective or Participle + (Qualifier) 273 

10.43. Omission of Headword in Binary Combinations 274 

10.431. Attributive Noun + (Noun) 274 

10.432. Adjective -f (Noun) 274 

10.433. (Noun + Prep.) -f Noun 275 

10.44. Omission in Longer Combinations 276 

10.5. A Note on F. Wellander's Theory of Ellipsis 277 

10.51. Introductory Remarks 277 

10.52. Definition of Ellipsis 277 




Partial Ellipsis 277 

Total Ellipsis 278 

Partial Ellipsis !> Total Ellipsis? 279 

The Parallelism 280 

Conclusions 281 

Ch. XI. Glass IV. Sense-changes due to Nomination. 

ii.i. Theoretical Discussion 282 

ii.ii. General Characteristics 282 

1 1. 12. Types of Nomination 282 

11. 13. Intentional and Unintentional 284 

11. 14. Intentional or Singular? 287 

11. 15. Causes and Conditions of Nomination 288 

11.2. Intentional Naming 291 

J 1.3. Intentional Transfer (Non-figurative) 293 

11. 31. General Remarks 293 

11.32. Transfers based on Similarity 294 

11.33. Transfers based on Other Relations 295 

11.331. Proper Names for Objects 295 

11.332. Place Names for Products or Events 295 

11.333. Christian Names 296 

1 1.4. Figures of Speech 296 

1 1.4 1. Introductory, and Definition 296 

11.42. The Delimitation of Metaphors 298 

1 1.5. Metaphor (incl. Hyperbole and Litotes). Theoretical Discussion 301 

11. 51. The Psychic Process: The Speaker 301 

11.52. The Psychic Process: The Hearer 304 

11.53. The Point of Similarity 305 

V' 11-54. The Metaphorical Experience 307 

11.55. The Relation between the Referents 309 

11.56. Hyperbole 310 

11.57. Litotes 312 

1/ 11.58. The Classification of Metaphors 314 

11.59. A Note on Marty's Theory of Metaphors 315 

11. 6. Metaphors based on Similarity 316 

1 1. 61. Nouns 316 

1 1. 61 1. Similarity of Appearance 316 

(a) Object for Object, Concrete 317 

(b) Object for Person 317 

(c) Proper Names in Appellative Use 317 

II. 612. Similarity of Quality, Activity, or Function 318 

(a) Object for Object, Concrete and Abstract 318 

(b) Object for Person 319 

(c) Quality for Person 319 

(d) Animal for Person 320 


(e) National Name in Appellative Use 320 

(f) Proper Name in Appellative Use 321 

(g) Place Name in Appellative Use 322 

1 1. 613. Similarity of Perceptual or Emotive Effect 322 

(a) Synaesthesia 322 

(b) Abusive Words as Endearments 323 

(c) Appreciative and Depreciative Uses 325 

11.62. Adjectives 325 

11.63. Verbs 326 

1 1.7. Metaphors based on Other Relations 326 

11. 71. Dress, Tool, Implement, etc., for Person 327 

11.72. Symbol for Thing Symbolized 327 

11.73. Material for Object 327 

11.74. Proper Names in Appellative Use: for Persons, Animals, or Objects 328 

11.75. Dates for Events 329 

11.76. Habitual Expression for Person 329 

11.77. Place Names in Various Uses 329 

11.78. Irrational Metaphors 330 

11. 8. Euphemism 330 

1 1. 81. Definition. Formal Types 330 

11.82. Causes of Euphemism 332 

11.83. The Psychic Process 334 

11.84. The Sense-change 336 

1 1.9. Irony 336 

Ch. xn. Class V. Transfer. 

12. 1. Theoretical Discussion 340 

1 2 . 1 1 . Analysis 340 

12.12 Conditions and Causes 343 

12.13 Delimitation and Definition 345 

12.14. Classification 346 

12.2. Nouns 347 

12.21. Identity of Appearance 347 

12.22. Identical Function 347 

12.23. Identity of Relative Situation within a larger Whole 348 

12.3. Adjectives 348 

12.31. Identical Appearance, Form, or Structure 348 

12.32. Identical Function, Ability, or Behaviour 349 

12.33. Relational Shifts 349 

Ch. XIII. Class VI. Permutation. 

13. 1. Theoretical Discussion 351 

13. 11. Analysis 351 

13.12. Conditions. The Equivocation 355 


13.13. Causes 359 

13.14. Delimitation and Definition 360 

13.15. Classification 361 

13.2. Nouns 362 

13.21. Objects' Names (Concrete and Abstract) 362 

13. 211. Material for Object made from it 362 

13.212. Receptacle for Content 363 

13.213. Part or Constituent Detail for the Whole, and Vice Versa 363 

13.214. Symbol for Thing symbolized 365 

13.215. Instrument for Action 365 

13.216. Instrument for Product 36^ 

13.217. Organ or Object for Capability of Perception or Intellection 365 

13.218. Articles of Dress or Equipment for Person 367 

13.219. Name from Concomitant Circumstance 368 

13.22. Nomina Actionis 369 

13.221. Action for Product, Result, or Object 369 

13.222. Action for Instrument or Means of Action 371 

13.223. Action for Agent 371 

13.224. Action for Place of Action 372 

13.23. Names of Qualities in Various Uses 372 

13.24. Names of Persons for Products etc 372 

13.25. Place-names 373 

13.251. Place-name for Action or Event 373 

13.252. Place-name for Inhabitants or Frequenters 374 

13.26. Mental State for Object or Person Causing it 375 

13.3. Adjectives and Adverbs 376 

13.4. Verbs 377 

13.5. Particles , 378 

Ch. XIV. Class Vn. Adequation. 

14.1. Theoretical Discussion 380 

14.11. Analysis 380 

14.12. Types of Adequation 382 

14.13. Conditions and Causes 385. 

14.14. Delimitation and Definition 385 

14.15. The Theory of Mechanization 387 

14.16. Classification 389 

14.2. Adequation after Substitution 389 

14.3. Adequation after Analogy 389 

14.4. Adequation after Shortening 390 

14.5. Adequation after Nomination 390 

14.51. After Intentional Naming and Intentional Transfer 390 

14.52. The Fading of Metaphors 390 

14.53. The Adequation of Hyperboles 393 

14.531. Cognitive Adequation 393 


14.532. Emotive Adequation (Fading) 396 

14.54. After Litotes 400 

14.55. After Euphemism 400 

14.56. After Irony 402 

14.6. Adequation after Transfer 403 

14.7. Adequation after Permutation 403 

14.8. Adequation without previous Sense-change 404 




Introductory Remarks 404 

Generalization 405 

Specialization (The Pregnant Use): Appreciative 408 

Specialization (The Pregnant Use) : Depreciative 411 

Particularization. The Unique Use 415 

Bibliography 421 

Index 433 


NED A Nexu English Dictionary, Oxford 1888 — 1928. 

COD The Concise Oxford Dictionary, 2nd ed. Oxford 1929. 

AF Anglo-French ME Middle English 

E English MHG Middle High German 

F French OE Old English 

G German OF Old French 

1/ Latin 


p. 32, last line: expressive signs; read: symbols. 

P. 33, line i: expression; read: symbol. 

P. 33, » 2: expressed; read: symbolized. 

P. 274, note i: einen Bock trinken; read: eine Bock ranchcn. 

and Change 
of Meaning 


1.1. The Position of Semasiology in Relation to Linguistics and 
Psychology. Semasiology is the study of linguistic meanings. The 
term meanings is here taken as denoting certain elements of the mental 
events occurring in connection with words in the minds of people think- 
ing, speaking, writing, hearing or reading the words. The study of 
meanings, as of all psychic phenomena, belongs to psychology. 

Meanings are accessible to scientific investigation only through the 
words which serve to express them, and they are linked up with these 
words in a manner that brings them into the province of linguistics, 
and makes their study an indispensable complement to the study of 
speech sounds and forms (cf. 3.25). 

Semasiology is thus a t5^ical Grenzwissenschaft, a boundary science, 
drawing material and principles both from linguistics and psychology, 
and to some extent also from epistemology and logic. In any single 
investigation, the emphasis may be more on the Unguistic side or more 
on the psychological side of the problem. 

The present work is a contribution to the estabhshing of semasio- 
logy as a branch of linguistics. Psycholog}^ and epistemology and 
logic, are placed under contribution as far as is necessary. 

1.2. Methods of Work. In an enormous field of work like semasio- 
logy, no single method can be the one and only way of salvation, but a 
number of different methods have to be used, according to the material 
handled and the end in view. In order to show the position of the pre- 
sent treatise within the whole, I shall mention briefly the principal 
lines of work.^) 

One distinction has already been mentioned: that between linguistic 
and psychological points of view. This distinction is crossed by that 
between descriptive and genetic methods; the two latter often overlap. 

The descriptive method may be defined, in general terms, as the mak- 

^) For details I refer to Grundtvig's bibliography. 


ing of an inventory of some part of the meanings and means of expres- 
sion in any one language at a given period. An ordinary dictionary is 
such an inventory, arranged from the point of view of the word. Its 
usefulness for purely semantic purposes is limited by the alphabetical 

Word-lists like Roget's Thesaurus of EngUsh Words and Phrases 
{Begriffswdrterbucher) represent another type, based on similarity of 
notional content. A dictionary of synonyms generally shows a blend of 
alphabetical and notional arrangement. 

The referents (see on this term ch. 3) are the basis of research in 
onomasiology, which investigates the names of objects in one or more 
languages. The linguistic atlases of France, Germany, and Italy are 
notable instances. 

Yet another type takes as its basis stylistic, morphological, or syntac- 
tical peculiarities, and makes a study of, for instance, the metaphors, 
the analogical formations, or the shortened expressions of a language. 
Parts II and III of Wellander's Studien zum Bedeutungswandel im 
Deutschen are an instance, dealing with shortened expressions in 
Modern German. 

Descriptive work on the psychological side will attempt to analyse 
the psychological (and epistemological) nature of meaning, regarded as 
static, as well as the nature of the psychic processes involved in the 
thinking of meanings, in their production and comprehension, i. g., what 
Paul called die gewbhnliche Sprechtdtigkeit. The relations between mean- 
ing, word, and referent, form another important problem. 

Turning to questions of detail, different types of meaning that may 
be distinguished from a psychological or logical point of view, have to 
be analysed for their pecuHarities: phrase-meanings and word-mean- 
ings, basic and relational meanings, and so on (see ch. 4); also the 
peculiarities characterizing the meanings of metaphors, shortened ex- 
pressions, etc. 

The genetic method can be applied along two different lines. One 
of them consists in tracing the history of words with regard to changes 
in the notional content of their meanings — so-called sense-changes. A 
historical dictionary, like the New English Dictionary, is a collection of 
such word-histories, arranged in an alphabetical order. Studies of the 
history of single words or word-groups have been pubhshed in relative- 
ly large numbers, although the methods made use of are not always 


the best. This line of genetic research has attracted more attention 
than the other, owing to its importance for etymology and lexicography. 
What a semasiologist would wish to see, in this direction, are diction- 
aries arranged on the principles of Roget's Thesaurus, with statements 
on the origin of every meaning of each given word. Such a book 
would form the basis for a thorough study of the semantic history of 
a language. It would also be of the greatest benefit to etymologists, 
who would find noted in it information on all developments observed 
for every meaning, and on the circumstances in which each single sense- 
change is liable to take place (cf. Grundtvig 8). We are as yet a very 
long way from the realization of this pium desiderium. 

Secondly, the genetic method may be applied to the study of the 
psychic processes underlying the notional changes of meaning. It is 
possible to study the notional changes without taking into account the 
psychic processes through which they arise; but, on the other hand, it 
is not possible to study these processes without taking up the notional 
changes, since it is only through their results in the form of such changes 
that the activity of the psychic processes becomes apparent. Moreover, 
any thorough scientific study of sense-changes must naturally seek to 
trace them to their causes, and the two genetic methods are therefore 
both of them indispensable. In Swift, Swiftly, and their Synonyms, I 
have shown how they may be applied to a group of words. Since 
the status of any language at any period is a phase in its unceasing 
development, and the result of the antecedent changes, the genetic 
method is the key to a real understanding of the semantic phenomena- 
Collection and analysis of historical facts form an indispensable basis 
for a scientific study, but do not alone constitute a science: analysis 
must be combined with synthesis, a synthesis based on general prin- 
ciples. In the case of language, which shows a development in time, 
the principles must be genetic. Since language is a product of 
mental activity, and the study of such activities belongs to psychology, 
it is in psychology that the principles of semasiology must be sought. 
On the other hand, a scientific study must be based on ascertained 
phenomena, and these are primarily the meanings and changes of 
meaning that have actually occurred in the course of Hnguistic 
development. It follows that an adequate historical linguistic material 
is an indispensable basis for semasiology. 

The adequacy of a theory in explaining the actual events in the 


semantic history of a language is the most important criterium veri for 
the theory. There is, however, another test which must not be neglected. 
Some at least of our mental processes are accessible to experimental 
and statistical investigation, and the results of this research must be 
incorporated in the theory, in so far as they touch upon relevant mat- 
ters. A good instance is the production of "artificial" analogies by 
Esper, as reported in 9.21. 

The aim of the present essay is to establish a theoretically tenable 
and practically workable system of classification comprising all known 
t5^es of sense-change. This means that the genetic method must be 
applied. There must be a Unguistic material in the form of word- 
histories, showing the various types of sense-change, the Hnks in the 
process of development, and the circumstances which determine each 
phase. On the theoretical side, it is necessary to analyse in detail the 
epistemological and psychological nature of meaning, so that the theo- 
retical differences between the various classes of change can be clearly 
demonstrated. For the same purpose, it is also necessary to analyse 
the speech process — die gewohnliche Sprechtatigkeit — both on the 
emissive and on the receptive side. 

The problems involved in the actual application of these principles 
will be further discussed below. 

1.3. Previous Work. Systematic semasiology is a comparatively 
late branch of study. It was only natural that the scientific study of 
languages, which dates back barely a century, should first turn to speech 
sounds and forms, as being most easily accessible, taking up problems of 
meaning only when it was unavoidable. The accumulation of linguis- 
tic material in dictionaries and handbooks of all kinds has hitherto 
mainly been intended to serve the requirements of practical Ufe, phono- 
logy, morphology, or syntax, while the needs of semasiology have sel- 
dom been taken into consideration as a prime factor. Since an ade- 
quate material is the indispensable basis for any science, semasiology 
has suffered accordingly. 

The first attempts to classify the semantic changes had no satisfactory 
theoretical and material foundation. Moreover, the psychology of 
that period was not yet ready to cope with the relevant theoretical 
problems.^) Practically the only method available was to compare the 

^) The history of semantic research up to about 1900 is sketched by Jaberg, 
Pej. Bedeutungsentw. For the last decades I refer to Grundtvig's bibliography. 


meanings of a word before and after a change, with regard to their 
notional content. It was found that, from a logical point of view, the 
secondary (derived) meaning often represented an extension or a restric- 
tion (specialization) of the primary meaning, or else could be otherwise 
logically related to it. In criticism of this method it is sufficient to say 
that, as the following discussions will show, the logical point of view is 
not relevant to the genesis of sense-changes: it can describe the results 
of a change, but can say little about its nature, conditions, and causes. 
In addition, there are many sense-changes which cannot be thus 
classified, and a classification should be exhaustive. 

The first writer to make a systematic attempt on a large scale to 
apply modern psychological principles to the explanation of linguistic 
phenomena was Wilhelm Wundt, in Die Sprache (first pubHshed in 
1900). In the semantic part, he seeks to classify sense-changes according 
to the nature of the psychic processes involved in them, or causing them. 

It is not my task to give a criticism of Wundt's opinions concerning 
language (see on this point especially Biihler, Krise 30 sqq.). I shall 
only remark that Wundt came to the study of language with a ready- 
made psychological system, based on research in other material (cf. 
Cassirer, Phil. Ill 240). He wanted to show that this system was ap- 
plicable to the linguistic phenomena — and in general to the "social" 
psychological phenomena — and at the same time he wanted to find 
in language fresh material in support of the system. 

Wundt's work has been much criticized, and sometimes uncritically 
accepted. It contains numerous inconsistencies and untenable explana- 
tions, and its main thesis has long since been abandoned. Experimental 
psychology was at that time in its infancy, and the progress of the last 
thirty years has shown up many weak points. A philologist especially 
notices Wundt's inabiUty to analyse historical hnguistic material in a 
methodical way. One excuse for Wundt in this respect is the circum- 
stance that very Httle really useful material was available to him. 

With these reservations, it is only fair to Wundt to acknowledge that 
no previous writer, and probably no subsequent writer, has made 
anything like so substantial a contribution to semantic theory. 
Wundt's great experience in psychological matters enabled him to give 
many explanations of permanent value. He has also the great merit 
of having posed and formulated many problems, and of having greatly 
intensified the interest in semantic questions, both among psychologists 


and philologists. His influence has been greatest among the latter, many 
of whom have regarded his work as the expression of a consensus — in 
reaUty non-existent — among psychologists, and even as a final formu- 
lation, long after active research on these problems had turned into new 
channels. Die Sprache has had to submit to the fate that sometimes 
overtakes efforts to estabUsh scientific systems. They summarize the 
state of knowledge at the time of writing; but research does not rest con- 
tent with a system, it goes on to find fresh material and to elaborate 
fresh methods for a more penetrating analysis, and thus inevitably 
brings to light new facts and factors which wreck the system. Wundt's 
theory and classification did not acquire any practical importance, and 
no one has been able to use it for classifying sense-changes. 

Wundt's failure seems to have discouraged others from taking the 
matter up in earnest, for during the last thirty years no one has made 
public any system of semasiology worthy of serious consideration. 
Semantic work from 1900 to 1930 has been characterized by an astonish- 
ing and highly regrettable lack of contact and collaboration between 
psychologists and philologists, a circumstance from which both parties, 
but especially the latter, have suffered greatly. Many philologists 
writing on semantic theory seem still to be ignorant of the fact that 
since 1900 new psychological, epistemological, and logical theories and 
investigations have been published of the greatest importance for the 
study of meanings; the decisive break with the older doctrines occurred 
between 1900 and 1910. Although admitting the necessity for borrowing 
theories and explanations from psychology and logic, these writers 
content themselves with antiquated doctrines, or with popularized 
fragments of psychological theories. Others, who are more up-to-date 
in this respect, speculate endlessly on the theoretical problems without 
troubling to verify their assertions by applying them to linguistic facts. 
And I do not think that any one has attempted to utilize systematically 
the findings of experimental psychology. 

The result of this state of things is that, in spite of the fact that 
there has been of late a not inconsiderable interest in the stud}' of 
meanings, we have as yet no generally accepted system of fundamental 
principles on which further research could be based. The principles 
adopted by Wellander, Falk, Funke, Sperber, Hatzfeld, Weisgerber, 
Junker, Ipsen, to mention only a few, are irreconcilable on many points 
of essential importance. 


1.4. Necessity for Systematic Work. In such circumstances, some 
writers recommend, as the only safe method of procedure for the 
present, the restriction of semantic work to problems of detail, until 
sufficient material shall have been assembled to furnish a satis- 
factory basis for a synthesis. This is, in my opinion, going from bad to 
worse. If we study the papers pubUshed on the semantic history of 
single words or groups of words, we are conscious of a serious deficiency 
in them all. It may be stated as an axiom that we do not get a real 
grip of any detail if we are not able to place it in its due setting, to 
fit it into the system to which it belongs, and thus to see it against a 
background of more comprehensive facts and general principles. Other- 
wise we cannot know what points of view are fundamental, we cannot 
know in what direction we should especially try to draw conclusions, 
and we have only a hazy idea, or no idea at all, whether the pecuUari- 
ties we notice are confined to this detail, or if they are essential to a 
larger class, and should therefore be specially emphasized. 

The lack of systematic knowledge also makes it difficult or impos- 
sible to see where the largest gaps in our knowledge are to be found, 
and what problems are essential for further progress, and we are thus 
incapable of directing our work to the portions of the field where it 
would be most useful.^) 

I quite agree with those who assert that much more detailed re- 
search is necessary in semasiology: there are large tracts that have as 
yet scarcely been touched, and no single portion has received a satis- 
factory treatment in any language. On the other hand, it is equally 
necessary to try to form some idea of the "lie of the land" as a whole. 
The work on details cannot be arranged to the greatest advantage if 
we cannot fit them into a general scheme, and make them support 
and explain each other by being put into relation to each other and to 
general principles. As it is, work is sometimes published that is al- 
most useless for general and systematic purposes, simply because the 
author did not know, and could not know, how to bring out the really 

^) "It may be the part of prudence, but it is not really the part of wisdom, to 
wait until all the data are accumulated and the evidence all in, before we try to 
theorize. The only danger in speculation lies in mistaking our formulations for 
established facts . . . Moreover, and this is the essential point, the real usefulness 
of theory lies in the fact that it sets specific problems to research and directs at- 
tention to aspects and details which would otherwise remain unnoticed" (Laguna 8). 


essential features of his material. I believe that this fact is largely 
responsible for the smallness of the annual output in semasiology. 
More philologists would take up this interesting work if they could 
only find a comprehensive theory ready to their hand, and were 
spared the trouble of constructing one for themselves. 

1.5. Plan of the Work. The present work is an attempt to estabhsh 
a theory and a classification of sense-changes on genetic, psychological 
principles. The plan has been (i) to try to ascertain, as precisely as 
possible, on the basis of typical and well-authenticated instances, 
what has really happened to the meanings investigated; (2) to ex- 
plain, with the help of adequate psychological theories, the conditions, 
causes, and nature of these changes. 

Such an investigation can be expected to show (i) what types of 
change exist in language, especially in the Enghsh language, and (2) 
how these types are related to each other. 

The meanings studied are the basic meanings of single words (cf. 4. 
26). I have thus excluded relational meanings, and phrase meanings. 
Neither of these types has as yet been more than slightly touched by 
research, and the material for a review of them is almost entirely 
lacking. I have had to confine myself to describing some of their main 
characteristics, and showing their relations to basic meaning. 

The material of basic meanings at my disposal is somewhat unevenly 
distributed. Most writers have concentrated on nouns; verbs and 
adjectives (with adverbs) have also had a share of attention, while 
pronouns and particles require much additional research. A complete 
review of the whole field of sense-change will perhaps be possible in 
two or three generations, on condition that the present rate of work is 
much increased. 

My position is, naturally, that of a philologist, and my work is a 
contribution to the building up of semasiology as a branch of linguistics. 
Psychology and epistemology have had to be given a prominent place, 
especially in the first part of the work, but they are there as Hilfswis- 
senschaften. It is the linguistic material in the second part that is the 
basis of the whole book; the theories are adduced to explain it, and 
are admitted only in so far as they are useful in this respect and are 
not contradicted by the material. The real test of any linguistic theory 
is its confrontation with the facts of language. 

My principle throughout is to apply an empirical method, as far 


as this can be done at present, and the foundations of the work are 
(i) the Hnguistic material, which is to show what has happened to the 
meanings in the course of development, and (2) a system of representa- 
tive psychological theories for explaining and bringing into systematic 
order the facts ascertained. 

This programme is in the nature of an ideal which can at present be 
reahzed only in part. I presume that no one will want to quarrel with 
the principles; it all turns on their application. Some indications con- 
cerning the contents of the following chapters will show what I beheve 
to be the best method of procedure. 

Since there is no recognized semantic theory on which a classification 
of sense-changes could be based, I have been compelled to state, in Part 
I (chs. 2 — 6), the principles of which I have made use in the systematic 
part (Part II, chs. 7 — 14). I have Umited the theoretical statements, 
as far as possible, to matter that must be referred to later, with as 
much complementary material as was necessary to make the account 
a coherent whole. Many problems of great interest have therefore 
been passed over as not relevant to the present study. 

The first great problem for Part I is the definition of meaning.^) 
In order to provide an adequate background for the definition it was 
found necessary to make some preliminary statements concerning two 
questions. One of them is the functions of speech, which are intimately 
involved in the definition of meaning, as well as in the whole problem 
of sense-change. The other question is a general theory of signs. 
Words are signs, and in order to explain their peculiarities it was 
necessary to go into the characteristics of signs in general, showing how 
word-signs differ from other signs. These statements are found in ch. 2. 

According to the older view, meaning is a complex of images and 
emotive elements, associated with a word. This opinion has, however, 
been totally discarded by modern research, and meaning is instead 
regarded as a psychic act through which the word (the sign) is referred 
to that which it denotes (the referent, see 3.13). The factors that go 
to the creation of this psychic act, as well as the interrelation of these 

^) We find writers — who claim to be working scientifically — attempting 
to analyse sense-changes and establish systems of classification without first stat- 
ing what it is that changes, i. e., without first giving an adequate definition of 
meaning (cf. Stern, Litteris III 49 sqq.). It stands to reason that such a neglect 
of the most elementary methodical principles can only lead to unreliable resijts. 


factors, have to be analysed. It is especially the relation between 
word and meaning that requires a detailed discussion, since a number 
of important problems are connected with it. The discussion of these 
matters is contained in ch. 3, which ends with a definition of meaning, 
formulated according to the principles previously estabhshed. 

Verbal meaning can be analysed with regard to the psychic nature 
of the elements constituting it, or with regard to its notional content; 
that is to say, from a psychological or from a logical point of view. 
In the former case we can distinguish cognitive and emotive elements 
of meaning; the cognitive being either thoughts or images, central or 
peripheral elements, and clear or vague elements; and we have to 
take up the problem of discriminating the meaning of a word from 
other mental content occurring together with it, which is either the 
meaning of other words, or is not meaning at all. From a logical 
point of view we have to establish various distinctions, as for instance 
between general and particular meaning, between basic and relational 
meaning, and so forth — the importance of which has not always 
been sufficiently recognized. These matters are treated in ch. 4. 

The second main problem for Part I is the nature of the speech pro- 
cess — what Paul calls die gewohnliche Sprechtdtigkeit. It is necessary 
to analyse the processes involved in thinking (producing or compre- 
hending) the relevant meanings, that is to say in producing and com- 
prehending speech. Our knowledge of these things is as yet far from 
complete, but the number of investigations is quickly growing and a 
considerable amount of fresh data is available. Both normal ps5^cho- 
logy and pathology have contributed, and some of the most important 
theoretical and experimental work will be briefly summarized. The 
discussion demonstrates the extreme variability and flexibility of the 
speech processes, a characteristic that is of essential importance as 
faciUtating the change of meanings. This analysis is contained in chs. 
5 and 6. 

The systematic part of the book begins with a chapter on the prin- 
ciples of sense-change. On the basis of the theories advanced in Part I, 
a definition of sense-change is first given, together with remarks on 
some general problems. The second section of ch. 7 contains a list of 
the seven main classes of sense-change, with typical instances. I may 
be allowed to state that I did not arrive at these seven classes by 
theoretical argumentation. I began by collecting and analysing well- 


authenticated sense-changes, and I found that they could be arranged 
in seven classes. The next step was to find out if these seven classes 
were statistical types only, or if it was possible to arrange them so as 
to form an organic system that could be deduced from simple basic 
principles. I found that it was possible if I made use of the definition 
of meaning given in ch. 3, and the analyses in chs. 4 — 6. Since I had 
arrived at the definition on purely theoretical grounds, while the classes 
of change were established empirically, this correspondence was an 
encouraging circumstance. 

1.6. The Linguistic Material. As long as we believe that it is 
always possible to build up the semantic history of a word on the 
basis of the bare statement that the word once upon a time had the 
meaning X, and that at a later time it had the meaning Y - — so long 
will semasiology remain a happy hunting-ground of faddists. If 
semasiology is to take its due place and position as an independent 
branch of linguistics, it must be placed on a secure foundation in the 
shape of a strict method of research. 

There are certainly many sense-changes the nature of which can 
easily be inferred from a knowledge of the primary and the secondary 
meanings, but these are the very simple cases, and no reliable system 
was ever built up on very simple cases alone. The comphcated cases 
are the crucial test of a classification (cf. Weisweiler's ironical com- 
ments on the "methods" still current in many quarters, St. u. Aufg. 

Moreover, such fragmentary evidence cannot help us in finding out 
the causes and circumstances of the change. That is to say, guesses 
may be made, and have been copiously made, but there is only one 
safe and scientific method: to undertake the laborious task of tracing 
in detail the sense-development of as many words as possible, with the 
help of chronologically arranged series of instances, taken from the 
texts of different periods.^) 

^) "Ce qui fait que les etudes sur le developpement du sens des mots, malgre de 
nombreuses tentatives, n'ont pas encore abouti a une theorie complete, c'est 
qu'on a voulu deviner les faits et qu'on ne s'astreint pas a suivre I'histoire des 
mots, et a tirer de I'examen de cette histoire des principes fixes; or, nulle part 
moins qu'en semantique, on ne pent determiner a priori les conditions de produc- 
tion des phenomenes; car en aucune partie de la linguistique les conditions ne 
sont plus complexes, plus multiples et plus varices selon les cas". Meillet, Ling. 


It follows from this principle that pre-hterary developments are best 
left aside at first, since the actual circumstances of the change can 
never be established, and that research should be restricted to periods 
represented by written texts. After having gained stable results there, 
it will be possible to conclude ex analogia with regard to earlier changes. 

It is also necessary to establish as exactly as possible the chrono- 
logical sequence in which new meanings arise, so as to avoid the risk of 
incorrect explanations, and to come as close as possible to the actual 
contexts in which the shifts took place. The circumstances may change 
so that later instances do not provide reliable information.^) 

We must try to show, then, how and in what contexts a word is 
used in different senses and at different periods; to infer as exactly as 
possible the shade of meaning in each quotation; and especially to 
ascertain in what context new meanings first make their appearance. 
We may then hope to find the explanation of each new shade of mean- 
ing, and if not the cause of the change, at least the precise conditions 
and extent of its action. 

This involves following up all the meanings of a word in detail, 
since we can never know in advance if there has or has not been 
organic development; sense-changes may be effected by non-hnguistic 
causes, by changes in the manufacture or appearance of the referents, 
or by our increased knowledge of them; analogical influences may have 
deflected the development into new channels, and so on. It is often 
indispensable to investigate also the history of cognate words, espe- 
cially if they are identical or very similar in form, and mutual influ- 
ence is suspected. Foreign influence must be taken into consideration 
in the case of translations or other works based on foreign sources, as 
for instance large portions of the OE and ME literature. 

It is evident that the consideration of so many points of view makes 
a thorough investigation of the semantic history of a word or group of 
words a laborious undertaking. It is further evident that the investi- 
gation of a group of synonyms has a great advantage over the investi- 

^) Cf. Paul, Wiss. I^ex. 72: "Fiir die historische Entwicklung der verschiedenen 
Bedeutungen eines Wortes bilden natiirlich die Ermittelungen dariiber, wann und 
wo dieselben zuerst vorkommen, eine unentbehrliche Grundlage, ohne die man 
leicht fehl greifen kann". The point is illustrated by relevant instances in Stern, 
Swift 4 — 5. It evidently makes a considerable difference if the origin of a parti- 
cular meaning has to be explained on the basis or OE, or of ME, or of NE condi- 


gation of a single word, since in the former case we often find parallel 
developments, and a gap in the history of one word may be filled in 
on the basis of evidence from the others (cf. various instances of this 
in Stern, Swift). 

It has been impossible for me to make a complete collection of 
fresh material for the present work; I have had to make shift with 
the material already available in print, together with additional matter 
taken from the New EngUsh Dictionary. It is to be regretted that 
the harvest to be gathered from that great work is not so rich as might 
be expected. The editors of a dictionary always seek to formulate 
their definitions as clearly and unambiguously as possible, and among 
the quotations available to them they select those that show the sense 
defined as free as possible from implications of other senses. The in- 
termediate uses of words, showing the transition from one meaning to 
another, are therefore not always found in the NED. But they are 
of essential importance for historical semasiology, as they often 
serve as links in the semantic development. 

There are other points to be observed. The possibility of influence 
from foreign languages, or from cognate words in the same language, 
by sense-loan, is a trap for the unwary, and requires detailed research 
to be detected. The "big" words in a dictionary are generally to be 
avoided. If studied in detail, with copious material, they are both 
interesting and useful, but the instances printed in the NED are seldom 
sufficiently numerous to allow conclusions to be made with safety 
(See Stern, Swift 4). The "big" words are especially liable to sense- 
loans and cross-influences, since they tend to split up into several 
branches which may influence each other in various ways. 

Words of foreign origin are mostly unsuitable as instances, since it 
is often uncertain what senses they brought with them into the Eng- 
lish language, what senses they have subsequently acquired through 
internal English development, and what they may have borrowed from 
their foreign prototypes. 

The best type for my purpose are words of one line of development, 
where collateral influences from other senses of the same word are 
excluded, but which are sufficiently well illustrated by quotations to 
show their history clearly. Even with words of this type care must 
be taken not to accept as absolute the dates for the first appearance 
of each new meaning. It has naturally not been possible, in this 


respect, for the NED to reach a level of reliability sufficient for 
detailed semantic research (see Stern, Swift 5). 

With these reservations, the KED is an excellent source, and it is 
indispensable for verifying dates and statements in other works. I 
acknowledge here once for all my great indebtedness to the KED, from 
which most of the definitions of meanings quoted in the sequel have 
been taken. 

Perhaps the caution advocated here will appear excessive. In my 
opinion it is not. Semasiology has suffered so badly from superficiality 
and the popularizing tendency, that what we need is a resolute ap- 
plication of strict critical principles, breaking once for all with the shp- 
shod methods too often practised. Only in this way will it be possible 
to estabhsh the study of meanings on a sound basis. The danger of 
any laxity is magnified in semasiology by our ignorance. We have 
not, as in phonology, a universally accepted system into which any 
new facts must fit, and an array of principles by which they can be 
tested. If we discover, or believe we have discovered, new and re- 
markable facts, we have no recognized standards to which they must 
conform, or else be rejected; it is consequently of the greatest 
importance that the material basis for a conclusion shall be above 
reproach, and it is by adding fact to fact in this way that a rehable 
system may in the course of time be established. At present, 
however, the progress is regrettably slow. 

It stands to reason that in a first attempt Hke the present one I 
have not been able to comply on all occasions with these strict prm- 
ciples; many things are still left to conjecture, from lack of sufficient 
data. I venture to think, however, that no reasonable precaution has 
been neglected. I have endeavoured to find rehable typical instances 
for each class or type of change examined. These instances are 
intended to illustrate my points, not to prove them; that is a task for 
future special research. I should add that considerations of space 
have prevented me from printing illustrative quotations for the in- 
stances adduced, except where it was absolutely necessary, as for the 
permutations (ch. 13). 

1.7. The Psychological Theories. Final Remarks. About the be- 
ginning of the present century, a number of psychologists in Germany, 
France, and America turned their attention more especially to the 
study of the higher mental processes, that is to say, to the kind of men- 


tal material that is of most importance in the meanings of words. Ex- 
perimental psychology up to that time had worked mostly with non- 
sense syllables; now investigators began to make use of real words, 
and to trace the processes involved in the production and comprehen- 
sion of sentences. It is evident that results gained in this way, on 
the basis of actual linguistic material, will be of greater value to 
philology than the results of experiments with meaningless words. 
With due caution, various results can be more or less directly appHed 
to linguistic problems. A number of these researches will be quoted 
in the following chapters. 

Many philologists entertain an insuperable distrust of psychological 
theories as applied to the phenomena of language. They point to the 
incontrovertible fact that psychologists disagree violently and con- 
sistently on many essential questions, and that psychological theories 
succeed each other with startling rapidity. We may concede all this, 
and still maintain the necessity of utihzing psychology for semantic 
work. If we look more closely, we shall find that things are not quite 
so black as they are painted: it is possible to discriminate. There 
exists a large, and constantly growing, body of experimental work, 
the results of which, if it is carefully performed and cautiously inter- 
preted, stand practically independent of the changing general theories. 
It is mainly psychological work of this kind that is useful to the philo- 
logist, as I hope this book will show. 

Many writers on epistemological and logical problems have also sought 
to establish their science on a more empirical basis, seeking contact 
with psychology. Husserl's Logische Untersuchungen (1901, 2 ed. 
1913) is the earliest work of this kind that I have consulted. It con- 
tains an analysis of meaning founded on principles proposed by F. 
Brentano, and altogether different from those of Wundt. Husserl's 
views have been extensively taken up by later writers. A whole volume 
on semasiology was published by H. Gomperz in 1908, comprising, 
among other valuable contributions, the first adequate definition of 
meaning on modern lines. 

Work on the new lines, both in psychology and epistemology, 
has continued without cessation, and important contributions have 
been made from the three countries mentioned, as well as from Eng- 
land. I refer to the quotations throughout this book, and to the Bib- 

1 6 GUSTAF STERN 1. 7 

These circumstances are of the greatest importance for semantic 
work. They enable us to approach the problems armed with theories 
and explanations that represent a great advance on those available 
to Wundt and his contemporaries thirty years ago. The new theories 
and experimental results have, as far as I know, not yet been syste- 
matically utilized for the purposes of semasiology although they have 
been accessible in print for a number of years^). 

The new theories and facts must be taken up in semasiology. If 
a scholar is compelled to borrow theories or other material from a 
branch of study not his own, and in which, therefore, he is not able 
to produce original research, it is, I think, obligatory on him to ascer- 
tain what the leading men of that other branch are teaching now — 
not what they taught thirt}^ years ago. 

None can know better than myself the difficulties of such a task in 
the case of psychology. Psychological books are pouring out in an 
ever increasing stream, and the task of picking out what is valuable 
and worthy of utilization is too great for one; it is impossible to ac- 
quire familiarity with everything. In a study that is not one's own 
it is difficult to move with certainty; the risk of overlooking relevant 
facts, or of generalizing on too narrow a basis, is ever present. I offer 
no excuse for m}^ mistakes and omissions; I can only hope that my 
contribution will stimulate others to improve on what I have done. 
It seemed to me that the work had to be attempted, and it is, I 
think, a work that should be undertaken by philology. It is no easy 
thing for a philologist to select and apply the necessary psychological 
theories and facts, but it is probably still more difficult for a 
psychologist to work in linguistic material. It is not enough to be 
conversant with modern languages. Historical material is indispens- 
able, and presupposes the ability to work with at least one language 
at all periods, ancient and modern. Still more important is a method- 
ical training in the handling of linguistic material and books of 
reference, a training that is not easily acquired by outsiders. 

1) See Pick's and Biihler's Berichte iiber Sprachverstandnis, 1909, as well as 
Biihler's papers in ArchfdgPs. 9 and 12. Useful summaries of the new departures 
may be found in Pick's Agrammatische Sprachstorungen 19 13, in Stahlin's dis- 
sertation on metaphors, 1913, in Willwoll, Begriffsbildung, 1926, and in Biihler, 
Krise, 1927. 


2.1. Language and its Functions. 

2.11. Speech and the Language System. The words and meanings 
discussed in the present work are the words and meanings occurring in 
the concrete acts of producing and comprehending speech. I shall use the 
term speech to denote both the activity and its results; including, conse- 
quently, (i) on the emissive side, the actual enunciating or writing of 
the verbal signs, and the thinking of their meanings; and on the recep- 
tive side, the hearing or reading of the signs and their interpretation into 
meanings; (2) the signs enunciated or written, and the thoughts that 
are their meanings. 

The signs themselves, their production and perception, need only 
occasionally be touched upon, my subject-matter being the thinking 
[cogitatio) of their meanings, as well as these meanings regarded as 
thoughts [cogitatum). 

Each member of a speaking community has learned, during the first 
years of his life, how to use his mother tongue. He has acquired a stock 
of knowledge regarding the form and use of words, their meanings, and 
the rules for their combination into phrases. We may term this stock of 
knowledge the language system, and describe it 'as the product of speech 
habits. These habits are imposed on the individual by the fact that if 
he wishes to be understood by his fellow-speakers, he must conform to 
their conventions. Each speaker can know only what he has been able 
to learn from his individual experience, and since different persons have 
different opportunities and Uve in different surroundings, no two 
language systems are completely identical. 

Each speaker must conform primarily to tht usage of his own group. 
But since all groups within a speaking community are more or less in- 
terlinked, there exists a body of conventions common to all speakers 
of one language, that is to say, there is a similarity on many points 
between the individual language systems. We may speak of a general, 

Goteb. Hogsk. Arsskr. XXXVIII: i 

1 8 GUSTAF STERN 2. 1 1 

transsubjective, language system, comprising what is common to all 
speakers, or at least to the majority, but we must take care not to hypo- 
statize this abstraction into a concrete entity, as some writers are fond 
of doing. ^) No language system has a real existence except as a stock 
of knowledge in an individual mind, and each mind can harbour its own 
system only.^) 

The distinction between speech and language sj'stem is based on the 
same fundamental principles as the distinction between actual and 
lexical meaning (4.22). An actual meaning is an element of speech; 
a lexical meaning is an element of the language system. 

As a comprehensive term, covering both speech and language system, 
I shall use language. 

2.12. The Functions of Speech. I shall speak here only of constitu- 
tive functions, essential to fully developed human speech. 

For a long time, it has been a widely diffused opinion that the func- 
tion of communicating mental content is the primary function of speech, 
even if not the only one. This view was held by Paul, although he gives 
no explicit definition to that effect, and many philologists will be in- 
clined to agree with him. On the other hand, some psychologists deny 
that the communicative function is essential (see note p. 21 below). It 
seems to me that it is difficult to do without the communicative func- 
tion in some of the analyses in the present study, and I consequently 
assume that it is constitutive (cf. 3.25). 

The expressive function of speech was rather one-sidedly stressed by 
Wundt (see esp. I 43, II 651). According to his theory, speech is a 
movement of expression, Ausdrucksbewegung. Mental processes, espe- 
cially of an emotive kind, often cause involuntary movements of a 
reflex nature. If the movements are localized in the respiratory and 
articulatory organs they may lead to the production of sounds, which 
Wundt calls Lautgebdrden, vocal gestures, and which he regards as being 

^) See for instance Sechehaye, Structure 12: "lya langue en se faisant un voca- 
bulaire cree done avant tout des signes d'idees." (My italics). Nothing is gained, 
except confusion, by such terminology. 

^) De Saussure has made a similar distinction one of the main points of his Cours 
de Ivinguistique, hut has turned the matter upside down by making la parole 
(speech) accessory, accidental, and secondary in relation to la langue (the language 
system). See Saussure 25 sqq., 30, 37. A carefully considered criticism of the 
"social" school is given by Delacroix 47 sqq. Cf. also Delacroix, 1,'analyse psycho- 


in origin and nature closely parallel to mute gestures. We find them 
used by most animals (cf. 2.22 on signals) . This phonetic material, existing 
before speech, was taken into the service of the developing human in- 
telligence, which for expressing its more and more differentiated and 
well-defined content required more and more differentiated and numer- 
ous sounds, and the constant interaction of speech and intelligence 
finally led to fully developed languages. The expressive function is 
universally regarded as essential for speech, but we now rate its part 
lower than Wundt did (cf. Buhler, Satz 3; Laguna 10 sqq.). 

It was one of the earUest views of language that words are primarily 
names of things: the essential function of speech is that of naming. 
During the last fifty years this theory has again come to the front, 
although in a modified form. We speak now of the symbolic function; 
the word symbolizes its referent (see on this term 3.13). The word 
symbol is then used in a restricted sense, as denoting a sign which also 
names that for which it is a sign; table is the name of an object, red of a 
quality, run of an activity, over of a relation. The symbolic function is 
especially characteristic of unemotional descriptive style. For instance, 
the passage, "Swift's father died before his birth, and his mother was 
very poor", is intended primarily to symbolize a set of facts {Sachver- 
halt); each word is linked up with a more or less definite referent. Of 
such a phrase we may say that it is true or false, not through its rela- 
tion to speaker or hearer, but through its relation to facts (cf. Gomperz, 
Sinn 83). It is the symbolic aspect of language that makes it a subject 
for logical analysis. 

We have, finally, the purposive or effective function, the importance 
of which is now generally recognized. The purposive nature of speech 
was emphasized by Marty, who states that our primary intention when 
we express our thoughts is "eine entsprechende Beeinflussung des 
fremden Seelenlebens"; intentional speech is a form of activity the 
main purpose of which is to evoke in other persons certain psychic 
phenomena. In comparison with this intention the expression or indi- 
cation of the speaker's own mental processes appears only as a means 
(Marty, Unt. 22, 284, 384, 433 sqq., 463, 493; Marty, Schr. II. 1.69). 
Later writers lay more emphasis on the use of speech for influencing 
the acts of other persons than for influencing their mental processes. 
"Once we deliberately ask the question: What does speech do? What 
objective function does it perform in human life? — the answer is not far to 


seek, speech is the great medium through which human co-operation 
is brought about. It is the means by which the diverse activities of 
men are co-ordinated and correlated with each other for the attain- 
ment of common and reciprocal ends. Men do not speak simply to 
relieve their feelings or to air their views, but to awaken a response in 
their fellows and to influence their attitudes and acts. It is further the 
means by which men are brought into a new and momentous relation- 
ship with the external world, the very relationship which makes the 
world for them an objective order" (I^aguna 19, 241 sqq., Pillsbur^^- 
Meader 15, 107 sqq., Gardiner 152, Rignano 92 with quotation from J. 
Mill, Cassirer, Spr. u. M. 32; Phil. I 42 — 43, 55 sqq., 246, 254 sqq.; Paul- 
han, Rev. Phil. 104, 22 sqq.). 

As a matter of fact, the theory of an effective function is a practical 
and logical consequence of the theory of an expressive function. A pro-- 
ducer of signs naturally implies a receiver of signs; they are correlative 
notions (cf . Biihler, Krise 33) . We need only remember that the absence 
of an audience generally means absence of all stimulus to speech. In 
solitude most people are mute; if there is no receiver of signs — no 
audience — speech is of no use, since it can have no effect. This does not 
mean that the receiver is passive; the comprehension of an utterance 
demands incessant activity on the part of the hstener, and the effect 
of the utterance is perhaps best described as a regulating of the receiv- 
er's own mental processes (Biihler, Krise 43 speaks of Steuerung). 
The imperative is the extreme case; but the term "influence" {Beein- 
flussung) must be taken in a wide sense; the purpose in view may be 
merely to inform the Ustener of something; this is usual in narrative 
style, as in the sentence just quoted: "Swift's father died before his 
birth, and his mother was very poor". In other cases the purpose may 
be to convince the Ustener of something, to make him perform some 
action, to make him react emotionally in a specified way, and so on. 

The theory agrees with the view that speech requires what Biihler 
(Krise 42) calls die unenthehrliche Zweieinigkeit von Zeichengebe^ und 
Zeichenempf anger , and partly also with the opinions of de Saussure and 
his followers concerning language as a social phenomenon. (Cf. Cassi- 
rer, Phil. I 254). 

The effective function must, I think, be the most important of the 
primary functions, for if, for instance, the words I am writing here are 
to have the intended effect on the readers, they must (i) symbolize 


certain referents, i. e., the topic which I am discussing; (2) express my 
views of it, and (3) communicate this to the readers; only if these three 
functions are filled, can speech (4) perform its effective function, which 
is thus founded on the others. The interrelations of the functions are 
not known in detail.^) 

As far as I can see, no other function of speech is essential. 

2.13. The Functions of Words and the Functions of Speech. We have, 
I think, to make a distinction between the functions of words and the 
functions of speech. I am speaking all the time of words in speech, not 
of words as elements in the language system (cf. 4.22). 

1) Biihler (Satz i sqq., Krise 30 sqq.) reckons with three functions of language: 
Kundgabe (expression, or, as Gardiner translates it, Br. J. of Psych. 361, self- 
expression), Auslosung (release, Gardiner: demand for response), and Darstellung 
{Gardiner: description), corresponding approximately to what I have called the 
expressive, the purposive, and the symbolic functions (cf. Willwoll 10 and Stern, 
Kindersprache 124 sqq.). 

Ogden-Richards (357 sqq.) distinguish five functions of "language as a means 
of communication": (i) symbolization of reference, (ii) the expression of attitude 
to listener, (iii) the expression of attitude to referent, (iv) the promotion of effects 
intended, and (v) support of reference. The referent is the topic of speech (the 
"object"), and the reference the relation between symbol and referent. See fur- 
ther ch. 3. Of these factors, (i) corresponds to Darstellung, or the Nennfunkiion, 
(ii) and (iii) to Kundgabe, or expression, split up into what are no doubt its two 
main factors (see 4.15 1), and (iv) corresponds more or less to Auslosung. With 
regard to (v), I think we may leave it aside; the explanation given by Ogden- 
Richards is not entirely convincing. It seems to me to be a phase of the attitude 
to the referent. In general, Ogden-Richards define language as an instrument 
for the promotion of purposes (1. c. 21). 

Biihler does not mention the communicative function; possibly he includes it 
in the Auslosungsfunktion. I do not see that the symbolic or expressive func- 
tions are either more or less essentially involved in the Auslosungsfunktion, the 
purposive function, than communication is. As emphasized by Laguna in the 
passage quoted above, an essential phase of the purposive function is the co- 
ordination of human activities: the communicative aspect is clearly of prime 
importance here. Its importance is not lessened by language occurring "im ein- 
samen Seelenleben" (Husserl II. 1.35, Marty, Unt. 494 sqq.) as a necessary method 
of formulating mental content that would not otherwise be apprehended with 
anything like the same clearness. 

Ogden-Richards seem to set the communicative function apart in some way, 
speaking of the various functions of language "as a means of communication". 
They do not state exactly how they consider the communicative function to be 
related to the other functions. — On the history of linguistic theory in general, 
see Cassirer Phil. I 55 sqq. 


The word has three functions: (i) the communicative function, (2) 
the symboUc function, (3) the expressive function (cf. 2.12). The discus- 
sion concerning the definition of meaning (ch. 3) will show that all three 
are essential; in individual instances one or the other may predominate. 
I have not been able to find any more constitutive functions in the 
meanings of words. 

Speech, on the other hand, has for its main function the promotion of 
purposes. No trace of this function is found in the meaning of the 
single word, when analysed alone, and it should therefore probably be 
regarded as a peculiarity of the psychic complex that constitutes the 
meaning of a complete utterance. Only in the one-word sentence, the 
extreme case, where utterance and word coincide, can the word acquire 
a purposive function. The latter is no doubt intimately connected with 
the initial adjustment to the situation described in 5.4 below. Such 
an adjustment can evidently occur only in actual speech, not in regard 
to isolated words or sentences -(4. 22), which have no bearing on the 
facts of real life. 

It seems not improbable that we should consider the purposive func- 
tion as a phenomenon of the kind called by psychologists configura- 
tional quaUties {Gestaltqualitdten) , which are pecuhar to complex forma- 
tions, but disappear when these are analysed into their constituent 
elements (see further 4.27, 5.4, 6.38, and Pick, Sprachstor. 130 sqq., 
with numerous quotations). What are the smallest portions into which 
speech can be split up without losing the purposive function, is a problem 
that I have to leave aside as not relevant to the present study. 

The functions belonging to the constituent parts — the words — be- 
long also to the constituted whole — speech. The purposive function of 
speech is conditioned by the three verbal functions, and could not be 
effective if the words employed in speech did not (i) communicate to 
the hearer a set of facts (Sachverhalt) which they (2) symbolize, and of 
which they (3) express the speaker's opinion. 

A detailed discussion of these problems falls outside the scope of the 
present study, and I refer the reader to the authors already quoted 
(concerning the origin of language see especially Laguna). The view 
that speech is primarily an instrument for the promotion of purposes, or, 
from another point of view, a purposive action, eine Zweckhandlung, and 
that this function involves communication, reference (symbolization) , and 
expression, agrees better than any other with the known facts of sense- 


change, and affords a plausible explanation for them. The uses to 
which a tool is put, naturally react on the tool itself, especially in the 
case of a tool so sensitive and flexible as language. In a great number 
of cases we have therefore to regard sense-changes as successive phases in 
the attempt, intentional or not, to adapt language to the purposes for 
which speakers make use of it. 

According to the hypothesis adopted, a certain preponderance is 
given to the speaker's point of view, the productive aspect. It is, after 
all, the speaker who employs language for his purposes. On the other 
hand, since his purpose is to influence the hearer's mental life, he must 
take care to be understood by his listener, and this fact immediately 
imposes on the speaker's use of language, and on language itself, cer- 
tain very definite limitations: understanding is as important a factor 
for the development of language as_speaking is (Delacroix 70). Here, 
too, the purpose reacts on the instrument. 

2.2. Signs and Meaning in General. 

2.21. General Theory of Signs. A word is a kind of sign or sym- 
bol. How can one entity become the sign of another; how can it acquire 
meaning? I shall give a summary of the explanations found in Ogden- 
Richards (138 sqq.; cf. also Pillsbury-Meader 157 sqq., Cassirer, Phil. 
I 26 sqq.). 

"The effects upon the organism due to any sign, which may be any 
stimulus from without, or any process taking place within, depend 
upon the past history of the organism, both generally and in a more 
precise fashion. In a sense no doubt the whole past history is relevant; 
but there will be some among past events in that history which more 
directly determine the nature of the present agitation than others. 
Thus when we strike a match, the movements we make and the sound 
of the scrape are present stimuli. But the excitation which results is 
different from what it would be had we never struck matches before. 
Past strikings have left, in our organism, engrams (Semon's termino- 
logy), residual traces, which help to determine what the mental process 
will be. For instance, this mental process is among other things an 
awareness that we are striking a match. Apart from the engraphic 
action of similar previous situations we should have no such awareness. 
Suppose further that the awareness is accompanied by the expectation 
of a flame. This expectation again will be due to the engraphic action 


of situations in which the striking of a match has been followed by a 
flame. The expectation is the excitation of part of an engram complex, 
which is called up by a stimulus (the scrape) similar to a part only of the 
original stimulus- situation". 

A sign, then, according to this view, is always "a stimulus similar to 
some part of an original stimulus and sufficient to call up the engram 
(if the reader is doubtful about engrams he may read "to call up an 
excitation similar to that caused by the original stimulus") formed by 
that stimulus". The general law is that "when a context (this term is 
defined p. 25) has affected us in the past the recurrence of merely a 
part of the context will cause us to re-act in the way we re-acted be- 
fore". (Ogden-Richards 139 — 140). i) 

It is especially important to notice that when, in the instance given, 
we are expecting a flame, "the mental process which is the expectation 
is similar to processes which have been caused by flames in the past", 
and it is "directed to" flame. "A thought is directed to flame when it 
is similar in certain respects to thoughts which have been caused by 
flames" (Ogden-Richards 140 — 141). 'Direction' is here equivalent to 
reference, which is the term I shall make use of in the sequel. 

Ogden-Richards give another instance that may usefully be quoted. 
"There is a well-known dog in most books upon animal behaviour 
which, on hearing the dinner bell, runs, even from parts of the house 
quite out of reach of scents and savours, into the dining room, so as 
to be well placed should any kind thoughts towards him arise in the 
diners. Such a dog interprets the sound of the gong as a sign. How 
does this happen? We shall all agree about the answer; that it is 
through the dog's past experience. In this experience there have been 
so to speak recurrent clumps of events, and one such clump has been 
made up roughly as follows. Gong, savoury odour, longing contempla- 
tion of consumption of viands by diners, donations, gratification. Such 
a clump recurring from time to time is what will be hereafter spoken 
of as an external context. Now on a particular occasion the gong is 

^) This agrees with the usual formulation of the general law of association, or 
Komplexerganzung. See for instance Claparede 165, Selz I 105 sqq., 175, Bur- 
kamp 259, Helson 36, 354, quoting Koffka: "If A, B, C . . . once, or several times, 
have been present in experience as members of a configuration, and if one of them 
appears bearing its membership character, then the tendency is present for the 
whole structure to be completed, more or less fully and vividly". 


heard out of reach of savours. But thanks to past experience of gong 
sounds together with savours in the interpretative dog, this present 
gong sound gets into a peculiar relation to past gongs and savours, 
longings, etc., so that he acts in the sagacious manner described, and 
is in evidence at the meal. Now this set of mental events — his present 
hearing of the gong, his past hearings of similar sounds, his past savour- 
ings together with gongs, etc., and also his present mental process 
owing to which he runs into the dining room — such a set is what will 
hereafter be alluded to as a psychological context. A context of this 
sort may plainly recur as regards its more general features. It is also 
clear that the members of it may be indefinitely numerous and may 
be widely separated in time, and that it is through this separateness 
in time that such a psychological context is able to link together ex- 
ternal contexts, the recurrent clumps of experiences of the gong-savour 
kind above mentioned. In a similar fashion all learning by experience 
will illustrate the point that to be an act of interpretation is merely to 
be a pecuUar member of a psychological context of a certain kind; a 
psychological context being a recurrent set of mental events pecuharly 
related to one another so as to recur, as regards their main features, with 
partial uniformity" (1. c. 143 — 145; cf. also Poppelreuter 251, Dashiell 25). 
To say that one thing is the symbol or sign of another is thus, accord- 
ing to Ogden and Richards, equivalent to saying that the thoughts 
of these two things are members of the same psychological context; 
and it is an indispensable condition that such contexts should recur 
with partial uniformity. "To say that I recognize something before 
me as a strawberry and expect it to be luscious is to say that a present 
process in me belongs to a determinative psychological context together 
with certain past processes (past perceptions and consumptions of 
strawberries). These psychological contexts recur whenever we recog- 
nize or infer. Usually they Hnk up with (or form wider contexts with) 
external contexts in a peculiar fashion. When they do not we are said 
to have been mistaken. The simplest terminology in which this kind 
of linkage can be stated is that of signs. Behind all interpretation we 
have the fact that when part of an external context recurs in experience, 
this part is, through its Unkage with a member of some psychological 
context, i. e., of a causally connected group of mental events often 
widely separated in time, sometimes a sign of the rest of the external 
context" (1. c. 145 — 146). 


Primarily, then, for A to be a sign or symbol of B, is for the thought 
of A to 'be directed to' or 'to refer to' the thought of B. I shall make 
use of the latter term, speaking of the reference of the symbol to the 
thing symbolized, and calling the latter the referent (see on this term 

Further, to be a sign or symbol of something is to have meaning. 
The terms are to be taken in the widest sense. A door has meaning for 
us because we know what it can be used for: it can be opened to give 
access, or closed to exclude (cf. Bourdon 195, Dashiell 25). 

The use of signs for communication is founded on the fact that a 
hearer may make a reference "similar in all relevant respects to that 
made by the speaker. It is this which gives symbols their pecuUarity 
as signs. Thus a language transaction or a communication may be 
defined as a use of symbols in such a way that acts of reference occur 
in a hearer which are similar in all relevant respects to those which are 
symboUzed by them in the speaker" (Ogden-Richards 333).^) 

We have thus arrived at the result that to be a symbol or a sign is to 
involve a reference to another entity, to some thing symbolized, and 
that this is equivalent to having meaning. Such a reference is founded 
on the fact that the entities are members of recurring contexts. The 
possibiUty of communicating by signs is dependent on similar references 
being made by speaker and hearer. (Cf. also Willwoll 9, quoting 
Meinong) . 

2.22. Signals, Symbols, and Substitute Signs. Having given a 
general definition of signs, we now have to analyse the way in which 
words function as signs. What has been said in 2.13 concerning the 

1) Ogden-Richards use the terms in a slightly different sense: "A symbol as we 
have defined it symbolizes an act of reference; that is to say, among its causes 
in the speaker, together no doubt with desires to record and to communicate, 
and with attitudes assumed towards hearers, are acts of referring. Thus a symbol 
becomes when uttered, in virtue of being so caused, a sign to a hearer of an act 
of reference. But this act, except where difficulty in understanding occurs, is of 
little interest in itself, and the symbol is usually taken as a sign of what it stands 
for, namely that to which the reference which it symbolizes refers" (1. c. 332). 

2) Cf. Russell's definition (Mind 1920): "i) A sign is an occurrence which, through 
mnemic causation, has mnemic effects (not, in general, other effects) appropriate 
(from the point of view of the animal's instincts and desires) to some other ocur- 
rence or set of occurrences with which it is apt to be associated. 2) In such a 
case, the other occurrence or set of occurrences is the meaning of the occurrence 
which is a sign". See also Bradley's definition, I,ogic I 4. 


constituent functions does not apply to all so-called words. If a per- 
son bangs his head against a door, and cries oh! the exclamation is a 
sign of the pain, but it is not the name of the pain, in the way that 
table is the sign and name of an object. Expressed in the terms defined 
in the preceding paragraph, oh! is not a symbol of thfe pain; it is 
essentially a reflex movement in which the expressive function 
predominates; it is precisely what Wundt called a vocal gesture 
(Lautgebdrde) , and his theory of Ausdrucksbewegungen is applicable 
here (cf. 2.12). The sound can be replaced by a mute gesture or facial 
contortion, or may be accompanied by them. I shall call signs of this 
type signals, a term once used by Cassirer, distinguishing them from 
the symbols, or signs which are also names for their referents. 

Words like damn, hell, etc., are sometimes used as signals, in more 
or less automatic reaction to pain or other feeUngs; but they can also 
be used as symbols, as names for their referents. (Cf. Cassirer, Phil 
III 127). 

Biologically, the signals are very old; most animals are able, through 
signals in the form of mute or vocal gestures, to make clear to others 
their own state of mind, while the use of symbols Hes altogether out- 
side the range of animal intelligence, and, in fact, constitutes the essen- 
tial difference between the language of animals and that of human 
beings. The former has, as Biihler says (Krise ^y), one dimension less. 

Our highly developed languages still make constant use of signals 
alongside of the symbolic words. Here belong not only gestures of all 
kinds, but also intonation, the alternation of stronger and weaker 
stress, pauses, speed of enunciation, and other ways of indicating our 
subjective attitude to the topic of speech that is named by the symbolic 
words, as well as the melodic, rhythmical, and echoic effects made use 
of by poetry. Signals and symbols are both indispensable elements 
in our stock of means of expression, and it would in most cases be 
very difficult to discriminate what we express by signals and what we 
express by symbols, or, conversely, what the listener learns from our 
symbolic words and what he learns from the running accompaniment 
of signals.^) 

^) Cf. Stout II 192 sqq., Husserl II. 1.23 on Zeichen and Anzeichen, and II. 1.78 
on the distinction between what he terms "Ausdriicke, die das Gegenstandliche, 
das sie nennen (oder iiberhaupt bezeichnen) zugleich kundgeben, und solche, bei 
denen der genannte und dev kundgegebene Inhalt auseinandertreten" . In my ter- 


Words may function as signs in a third way, termed by Stout (II,. 
192 sqq.) substitute signs. "A substitute sign is a counter which takes 
the place of its meaning; so long as it fulfils its representative function,, 
it renders useless all reference to that which it represents. The counters 
are manipulated according to certain rules of operation, until a 
certain result is reached, which is then interpreted. The operator may 
be actually unable to interpret the intermediate steps. Algebraical 
and arithmetical symbols are to a great extent mere substitute signs. 
The same is true of the symbols employed in formal logic." A symbol 
is, as Stout points out, a means of thinking about its referent; when 
we use substitutes we do not think about that which they symboUze.^) 
It is characteristic of the substitutes that they completely lack the 
expressive function (cf. Cassirer, Phil. Ill 393). 

In the present study, words as used symbols are the sole topic. 
The two other kinds of signs will only occasionally be touched upon. 

minology, between expressions that are both symbols and signals for one referent^ 
and those that are symbols for one referent and signals for another. In the as- 
sertion twice two is four, the utterance is a symbol for a mathematical fact; on 
the other hand, it is also a signal to the hearer that the speaker believes in the 
truth of the assertion. A symbol for this belief would be, / believe that twice two- 
is four. The former assertion in this sentence may be false, and the latter true. 
Cf. also Marty, Unt. 490 sqq., Cassirer, Phil. Ill 375, WillwoU 7 sqq., Gomperz, 
Sinn 81 sqq., and the discussions below in 4.14, 4.15, and '5.21. 

^) On the mathematical signs, see Husserl II. 1.68, Schwarz 157, Gomperz II. 
1. 133, Fischer, Arch. 43, 39 sqq. 


3.1. The Factors of the Meaning Complex. 

3.11. The Three Factors. In all normal use of speech, in all ordi- 
nary discourse {in der gewohnlichen Sprechtdtigkeit, as H. Paul terms 
it), when words are being used as symbols, we have to reckon with 
the following factors (cf. Pfander 7): 

(i) A thinking and language-using subject (speaking, writing, 
hearing, reading). 

(2) The acts of thinking taking place in the mind of the subject. 

(3) The thoughts forming the content of these acts of thinking. 

(4) The word-forms to which the thoughts are associated and which 
express them, more or less completely. 

(5) The subject, the acts of thinking, and the thoughts are referred 
(in the widest sense of this term) to some object, which I shall 
call the referent. 

This applies both to the producer and the receiver of speech. 

The distinction between acts of thinking (the subjective thoughts, 
cogitatio, das Denken), and the content of these acts (the objective 
thoughts, cogitatum, das Gedachte), is of importance for some problems. 
For the present, I shall make use of the term mental content as com- 
prising both cogitatio and cogitatum. I shall further use the terms speaker 
and hearer as equivalent to producer and receiver of speech, respect- 
ively. It is often necessary to consider speaker and hearer separately. 

The subject, speaker or hearer, enters into the present argument only 
in his aspect of thinking the mental content, and need therefore not be 
otherwise included. We have to keep in mind, however, that the men- 
tal content is not an independent entity, but is an individual person's 
mental content. The two primary factors in all speech are (i) the 
thinking and speaking (or listening) subject, and (2) the things spoken 


of (the referent).^) Words and their mental content are the instruments 
used by the subject for certain purposes, just as they are the instru- 
ments through which the hearer grasps the speaker's intention. This 
follows as a corollary of the view that speech is an instrument for the 
promotion of purposes, for the purposes are naturally those of an indi- 
vidual who uses speech to promote them. 

The fact that words and meanings are secondary, conditioned, phe- 
nomena, has been neglected by most philologists writing on semantic 
theory, and also by earlier psychologists. They handle words and mean- 
ings as primary, independent, entities, a view that leads them into 
various mistakes. 

I shall make use of the habitual terminology, according to which the 
words denote, name or designate, the referents, and express the mental 
content, although it would be more correct to say that the subject de- 
notes, names, or designates the referents, and expresses his mental con- 
tent by means of the words. It will sometimes be necessary to empha- 
size this point in order to avoid mistakes in the analysis. 

By comprising cogitatio and cogitatum under the heading of mental 
content, and by excluding the subject as a separate factor, since he is 
represented by his mental content, we reduce the five factors of 
meaning to three: 

(i) the word, expressing the mental content and denoting the 

(2) the mental content, connected with the word, and involving a re- 
ference to the referent; 

(3) the referent.'^) 

Before entering upon the details of the problem of definition, it will 
be useful to make some comments on the terms word and referent, and 
especially to discuss the question whether meaning may be identical 
with word or referent, or have any element in common with either. 

1) "(Die Sprache) ist niemals Aussage schlechthin, sondern immer lebt in ihr 
zugleich ein Modus, eine individuelle Form des Sagens, in der das sprechende 
Subjekt sich selbst ausspricht. AUe lebendige Rede schliesst diese Doppelheit, 
diese Polaritat von Subjekt und Objekt in sich. In ihr wird nicht nur auf be- 
stimmte Sachverhalte hingedeutet, sondern in ihr pragt sich die Stellung des Sub- 
jekts zu diesen Sachverhalten aus." Cassirer, Phil. Ill 393. 

^) See on this matter also Gomperz II.1.61, Husserl II.1.31, Moore, Mono. 243, 
Ogden- Richards 14, Gardiner 354, and Meinong's formulation, quoted Willwoll 9. 


3.12. The Word. Meaning is a property peculiar not only to what 
we traditionally call words, but also to parts of words, e. g. the geni- 
tive ending in man's, and to groups of words, e. g. compounds or clauses. 
As a comprehensive term Noreen (Betr. 200) has proposed semem, formed 
on the analogy of the French terms phoneme and morpheme; Vendryes 
(85) has proposed semanteme, and C. S. Peirce, quoted by Ogden-Rich.- 
ards (438) has seme and pheme. None of these coinings seems to have 
gained the approval of English writers. For my purposes it does not 
seem necessary to adopt any of them, or to coin a fresh term, since I 
am treating only of the meanings of what we traditionally call words, 
except where I have to discuss the relations between the meanings of 
words and those of other linguistic formations. 

The word, as distinguished from the mental content that it expresses 
and from the referent that it names or denotes, is in this book always 
to be understood as a physical phenomenon, acoustic, visual, or tac- 
tile, together with the verbal images, visual, acoustic, or motor (graphic, 
articulatory) involved in the pronunciation, hearing, reading, writing, 
and even in the mere thinking of the word. It is not necessary for 
me to make any further distinctions in this matter, or to discuss the 
very complicated relations obtaining within the word-complex.^) 

A "word" without an associated mental content has no meaning, it 
is merely a flatus vocis. That is the case with a word from an un- 
known language. If the sound-complex is to be apprehended as mean- 
ing something, as significant speech, a mental content must accrue 
to it. (Gomperz II. i. 65, Husserl II. i. 32 — 33, Martinak 2, Schmitt 
207, quoting Humboldt, Pos 103). 

3.13. The Referent. The referent is that which is denoted by a 
word, that to which word and meaning refer. It may be concrete or 
abstract, actually existing or imaginary; in short, anjrthing that is 
capable of being made the topic of formulated thought and speech. 2) 

Philologists have, as previously remarked, generally neglected to 
take the referent into account in their analyses of meaning and change 
of meaning. The necessity of doing so will be abundantly evident from 

^) See further Noreen, Betr. 203 sqq., 433, Jespersen, Phil, of Gr. 92, B. Erd- 
mann II. 362 sqq., Wundt I 568 sqq., Otto 36, Gutzmann 47, Moore, Mono. 100 

— lOI. 

2) The term referent is adopted from Ogden-Richards, who define it as "what- 
ever we may be thinking of or referring to" (Ogden-Richards 13, note). 


the following discussions. In all ordinary discourse our attention is 
directed towards the topic, the referents; our mind goes on at once 
from the words to the referents, from the symbol to that which it sym- 
boHzes, and promptly forgets all about the symbols, to which only so 
much attention is given as is necessary to deduce from them the 
reference to the topic (Cf. Schiller 385 sqq., Marty, Unt. 491, Feldkeller 
290, Pos 118). 

3.14. Word and Meaning. Can the word — in which term, as pre- 
viously stated, I include the verbal images — be wholly or in part 
identical with meaning? 

Such an opinion is held by the behaviourists, according to whom 
thought is the action of language mechanisms, and thinking is subvocal 
talking (J. Watson, Psychology from the Standpoint of a Behaviorist, 
14 and 316, quoted by Otis 399; Watson, Br. J. of Psych. 9; for 
criticisms and suggestions on other lines, see four other papers in the 
same volume; R. H. Wheeler, Development). A different view is de- 
fended by Marty, Unt. 459 — 460, Husserl II. i. 68, Gomperz II. i. 63, 
Spearman 181, Schiller, Mind 385 sqq., James I 251, 253, J. S. Mill, 
lyOgic I 196, quoted by Rignano 281, Moore, Mono. 85, loi, 174, 
224, 229. 

Gomperz points out that word and meaning may vary independently 
of each other, and therefore cannot be assumed to coincide. The 
same meaning can be expressed through different words, as German 
hreit and English broad; conversely, different meanings can be expressed 
by the same word, as instanced by crown or any other polysemous 

It is true that word and meaning are intimately fused (see 3.21), 
and also that the trains of thought characteristic of an educated adult 
person would not be possible for anyone not acquainted with the use 
of words, although such thought may nevertheless occur without them 
(Stout II 186). In spite of this fusion of word and meaning, Spearman 
shows that his observers were always able to distinguish the two 

It is the latter of the two opinions that agrees with the general po- 
sition in matters of psychology that I have taken up, and which must 
therefore be adopted for the purposes of this study. I assume that 
word and meaning never coincide. This seems to be the reasonable 
view when we apprehend words as expressive signs; the same entity 


could not very well be simultaneously expression and that which is 

3.15. Referent and Meaning. We have next to discuss the question 
whether meaning and referent can coincide, wholly or in part. 

When the referent is a material object, it can evidently not coincide 
with meaning, which is a psychic entity. 

With regard to abstract referents, opinion among psychologists is 
divided. Stumpf contends, according to Titchener (Am. J. of Ps. 33, 
48) that when our thought is directed upon the universal as such, upon 
concept or law, then content and object (approximately equivalent to 
what I call meaning and referent) coincide: the content is, by its very 
nature, object; or, in my terminology, the objective thought (cogitatum) 
is identical with the referent thought about. Thus, for instance, the con- 
cept of relativity, when spoken of, would be both meaning and referent. 

We have here perhaps a difference in the interpretation of terms. 
When anyone speaks of relativity, the meaning of the word, according 
to the definition that will be given below, is the speaker's subjective 
apprehension of the concept of relativity {i. e., of the referent), and it is 
clear that such apprehension will vary widely for different individuals, 
as well as for the same individual on different occasions. But the trans- 
subjective concept of relativity remains untouched by these variations. 
I therefore make a strict distinction between the concept of relativity 
as referent, and the various individual ways of apprehending this re- 
ferent, as meanings. This view is supported by Husserl II. i. 43, 46, 
370 sqq., and Honigswald 131 — 132, quoting Bolzano's Wissenschafts- 
lehre (of. 4.23). 

-"Gomperz (II. 1.63) insists that Aussagelaute and Aussagegrundlage 
cannot coincide because very different referents can be apprehended 
through the same meaning, and different meanings can refer to the 
same referent. ^ A meaning expressed by the sentence There flies a bird 
can refer to a fluttering sparrow, to an eagle or a swallow flying along, 
etc. On the other hand, a sparrow fluttering outside the window may 
be referred to by meanings expressed in sentences Uke there flies a bird, 
there is a sparrow fluttering, look, there is an animal, how frightened it is, 
€tc. These may all be correct apprehensions of the same referent. 

I think the simplest way is to follow Husserl and Gomperz. Their 
theory gives us the same distinctions between word, meaning and re- 
ferent throughout the whole field, whether the referent is concrete or 

Goteb. Hogsk. Arsskr. XXXVIII: i 

34 GUSTAF STERN 3. 1 5 

abstract. Especially with regard to abstract referents, it seems useful 
to distinguish clearly the referents, on the one hand, and the meanings 
through which the referents are apprehended, on the other. 

3.16. Mental Content and Meaning. If meaning does not, even in 
part, coincide with word or referent, it follows that it must be sought 
entirely in the mental content that accompanies speech. We have to 
consider whether meaning is identical with the mental content as a 
whole, or with certain elements of it. 

It is easy to see that meaning cannot be identical with the total men- 
tal content occurring in connection with a word, or present in conscious- 
ness simultaneously with it. We are, for instance, always more or less 
conscious of ourselves and our surroundings. As I write, I am con- 
scious of the paper and the typewriter, of the table in front of me and 
of the hght from above. I hear sounds from an adjoining room and 
from the street, and I feel the current of air from the open window. 
The consciousness of these concomitant cirumstances lies at the back 
of my mind, forming a dimmer fringe to the focal elements on which 
I am concentrating my attention. It is evident that this consciousness 
does not belong to the meaning of the words I am using, except when, 
as just now, I am making it the topic of discourse. This appUes to 
both speaker and hearer (Dewey 6, Moore, Mono. 190 c). 

A feeling of familiarity or recognition may also accompany the per- 
ception of a word, although it is generally not noticed, but merged in 
the meaning. However, it may occur in isolation, when we recognize 
a word, but do not remember what it means. This feeling is connected 
with the word, but does not belong to its meaning. See 6.22 below. 

We have further to note that associative processes originating in a 
word may go in other directions than, and far beyond, its meaning. 
Harmodios may evoke the thought of Aristogeiton, but Harmodios does 
not mean Aristogeiton (Gomperz II. i. 139. Messer, Arch. 61). 

A person speaking may have his mind at work selecting and judging 
the material he is going to present to his listeners, weighing it as he goes 
along, and performing other mental operations which do not enter into 
the meanings of the words used, although they are intimately connected 
with them and may have been decisive for the selection. 

A Hstener, too, does not merely receive and passively register the 
meanings of the words he hears, but proceeds at once to combine them 
with what he previously knows or has heard of the topic, to form judge- 


ments, approve or reject, and so on (cf. 6.33). All this is mental con- 
tent connected with the phrases heard, but only part of it can be said 
to belong to the meanings of the words. The rest I shall call mental 
context. (See 4.16 and 6.32). 

It is evident enough from these brief indications that meaning is not 
identical with all the mental content associated with the word, but 
only with part of it. Mental content thus forms the genus proximum 
for a definition of meaning. It remains to find an adequate differentia 
specif ica. 

3.2. The Definition of Meaning. 

3.21. The Connection between Word and Meaning. Although mean- 
ing has not yet been precisely defined, we have arrived at the con- 
clusion that it must be a portion of the mental content connected with 
the word. Before proceeding to the definition it is necessary to ana- 
lyse the nature of the connection. 

According to the older view — still represented by some writers — 
the connection between a word and its meaning is nothing more than 
an association of contiguity; this view, says Biihler (Arch. 12, 85) must 
be regarded as "geradezu naiv". 

Recent investigators seem to be unanimo-us in describing the connec- 
tion between the word (as a symbol) and its meaning, the meaning 
relation, as a relation sui generis, which cannot be analysed into com- 
ponent elements. "Bezeichnen wir die Beziehung, derzufolge ein 
Sinnliches einen Sinn in sich fasst, und ihn fiir das Bewusstsein un- 
mittelbar darstellt, als die der "symbolischen Pragnanz", so lasst sich 
der Sachverhalt dieser Pragnanz weder auf bloss reproduktive noch 
auf mittelbare intellektuelle Prozesse zuriickfiihren: er muss zuletzt 
als eine selbstandige und autonome Bestimmung anerkannt werden". 
(Cassirer, Phil. Ill, 273). Ach has made extensive experiments in 
elucidation of the problem (Ach, Begriffsbildung; see also the summary 
in Willwoll ig sqq.). His conclusion is, briefly, that the meaning rela- 
tion comes about on the basis of a causal connection or an association, 
but that these alone can only lead to the word becoming a signal for 
another item, carrying what Ach calls indizierende Bedeutung. In 
order to be apprehended as a symbol, to receive signijikative Bedeutung, 
there must be added a new factor, through which the word is made to 
represent its referent in quite a new way. On the basis of his reports. 

36 GUSTAF STERN 3. 2 1 

Ach describes das Erlebnis der Bedeutungsverleihung as "ein plotzlich 
einsetzendes, wie eine Erleuclitung wirkendes Erlebnis, das durch das 
Auftreten, bzw. durch das Klarwerdeti des inneren Zusammenhanges 
zwischen dem Vorwort (the word that is apprehended as a name). . . 
und der Eigenart der Nachworter (the group of words that are appre- 
hended as referents) gekennzeichnet ist, oder durch die Identifizierung 
zwischen dem Vorwort und der Objektvorstellung der zugehorigen 
Nachworter, und das von lebhaften Lustgefiihlen begleitet ist" (Ach, 
Begr. 196, also quoted Willwoll 25). The attribution of meaning to a 
Vorwort may happen under pressure of a task which the observer can 
best solve by employing the Vorwort to denote the group of Nachworter 
Secondly, it may come about through a Benennungsurteil, as when we 
are told, "that is a gostok" ; and, thirdly, as the result of latente signifi- 
kative Einstellimg, our latent tendency to give names to everything. 
Through any of the three processes, the intrinsic connection as distinct 
from the exterior association, between significans and significatum is 
suddenly realized, and a fusion [Verschmelzung) takes place between 
them. The result is a complex that Ach calls jusional unit {Fusions- 
einheit), and through this fusional unit the word (the symbol) "wird zum 
anschaulichen und unmittelbaren Trager der Objektvorstellung, und 
bedeutet so deren Gegenstand" (Ach 299 — 300). Similar results of 
other experiments are described by Spearman (211). The name is 
then the means of mutual understanding, and of denoting, in an 
unequivocal manner, a fact or set of facts {Sachverhalt). It is necessary 
that the person addressed should have experiences that enable him 
to understand what the speaker means, that is to say, he should 
possess similar concepts. 

By virtue of the fusion thus described the name also becomes a 
characteristic of the referent. If some large and heavy cardboard ob- 
jects are called gazun, then the being gazun is a characteristic of these 
objects in exactly the same degree as the being large and the being 
heavy. And the being gazun, i. e. the name gazun, is even the most 
important characteristic of the referent, in so far as we are concerned 
with this name as a means of communication. All other characteristics 
run the risk of being found unessential in the course of development, 
but the name never (1. c. 307). This agrees with Wundt's analysis 
(II 510— 511). 

Ach's experimental arrangements were artificial in so far as the name 




was provided. In ordinary discourse the name of a previously un- 
named referent has to be found by the speaker, and he generally se- 
lects a characteristic of the referent which for some reason is pro- 
minent in attention, and applies to it a name that he finds suitable, 
invented for the purpose or transferred from another referent.^) 

3.22. The Interrelation of the Three Factors. In order to show 
clearly the interrelations of the three factors of meaning, I reproduce, 
with some modifications, the diagram given by Ogden-Richards (p. 14). 
A similar diagram is given by Gomperz II. i. 77. The terms in brac- 
kets are from Ogden-Richards. 

(Thought or Reference) 


Denotes or Names 
(Stands for, an imputed relation) 


1) These matters have been the subject of much discussion. I may give the 
following references: Biihler, Bericht 112, Messer, Arch. 175 — 176, Gomperz II. 
1. 132 sqq., and 258.1, Maier 248, 342 sqq., Marty, Schr.II.1.67 sqq., Pos 37 
Eqq., 59 sqq., Fischer, Arch. 42, 352 and 43, 37, Selz II 380 sqq., Martinak 9, 
49, Schwarz 158, Stern, Psych. 132 and Kindersprache 190 sqq., describing the 
awakening of the Symbolbewusstsein in children. Ach himself refers to Herbart, 
Wundt, Stumpf, P. Krueger and H. Maier. See especially Cassirer III 131 sqq. 


We have already discussed the relation holding between word 
and meaning (symbol and thought or reference); that between mean- 
ing and referent is treated below. The third relation, that between 
word (symbol) and referent, is clearly not a direct relation: the word 
is referred to the referent by the speaker (hearer) and the relation 
is thus mediated by his thought.^) 

Instead of saying with Ogden-Richards that the symbol "stands for" 
the referent, and symbolizes the thoughts, I shall make use of the 
terms already indicated above (3.11): that the word expresses the 
mental content (meaning, thought) and names or denotes the referent. 
The mental content is the speaker's apprehension of the referent. 

In the following discussion, I confine myself for the present to the 
cognitive (noetic) elements of meaning, leaving aside the emotional, 
"non-symbolic" (Ogden-Richards 13) elements, which offer problems 
of their own, and will be dealt with later (4.15). 

3.23. The Objective Reference. It has been stated as a fundamental 
principle that the meaning relation between a symbol and the thing 
symbolized involves a reference to the latter, to the referent. It is a 
constitutive characteristic of all mental content that it refers to some- 
thing: a thought is always a thought about something. "We say of 
man, mouse, or monkey, that it feels, remembers, perceives, infers, 
desires, strives, and so forth. Leaving aside the first term, it is ob- 
vious that all the rest' imply both an activity and an object" (Ward 
60). "No idea is a "mere" idea, the cognitive function, i. e., the re- 
lation to something which it means, is essential to its very being" 
(Hoernle 76). 

In the passage just quoted Ward made an exception for feeling, but 
later on we find him stating that "emotion is always the expression of 
feeling, and feeling, for the subject that feels, has always some ob- 
jective ground" (Ward 275). Burkamp (258) thinks that only the pure 
hedalgedonic feeUng might be non-objective, for everj^thing else has a 
reference to that which is thought. And even the hedalgedonic feeling 
will be apprehended as referring to an object. We feel pleasure or 

1) Die . . . zwischen den A ussagelauten und der A ussagegrundlage stattf indende 
Relation der Bezeichnung ist lediglich eine vermittelte und rein ausserliche, und 
ihr Weseu besteht nur darin, dass aus dem Vorhandensein der Aussagelaute auf 
das Vorhandensein der Aussagegrundlage geschlossen warden kann. Gomperz 
II. 1. 132. 


displeasure at something, and it is only when this relation makes itself 
evident that the feeling is able to fulfil its regulative function in the 
psychic process. Shand (64) states that every emotion comprises also 
a cognitive attitude, and a conative attitude in the sense of an im- 
pulse and an end (see 1. c. 272). The matter is, however, controversial. 

The peculiarity of mental content to be directed, or to refer, to 
something outside itself should, in strict consistency, be called the re- 
ferential reference, but for obvious reasons I prefer to call it the ob- 
jective reference. 

The objective reference is, according to the principles stated, the one 
element of meaning that cannot disappear. However vague and evanes- 
cent the thought connected with a word is, there must, if the word 
is understood, exist an objective reference, since a symbol is not a 
symbol if it does not refer to a thing symbolized.^) 

The constitutive importance of the objective reference is also brought 
out by the fact that two persons may without difficulty converse about 

^) Among authors who have discussed this point I refer to Humboldt, quoted 
Ach 340, Stout I 47 and 122, II i, Dewey 75, Maier 152, Husserl II. 1.46, 54, 
363 sqq., 415 sqq., Marty Unt. 226, 423, 431, Marty Schr. II. 1.69, Pfander 7, 
Bosanquet I 38, Biihler, Krise 67, and especially Cassirer, Phil. III. 227: "Die 
moderne Phanomenologie hat in ihrer Definition und in ihrer Analyse der 
Wahmehmung nicht sowohl an Kant, als vielmehr an Brentano iind an seine 
Begriffsbestimmung des Bewusstseins angekniipft. Brentanos "Psychologie vom 
empirischen Standpunkt" findet das auszeichnende Moment des Bewusstseins, 
des "Psychischen" iiberhaupt, im Charakter des "Intentionalen". Ein Inhalt 
ist "psychischer" Inhalt, so fern er eine eigentiimliche Richtvmgsbestimmung, 
eine Bestimmung des "Meinens" in sich fasst. "Jedes psychische Phanomen ist 
durch das charakterisiert, was die Scholastiker des Mittelalters die intentionale 
(auch wohl mentale) Inexistenz eines Gegenstandes genannt haben, und was wit, 
obwohl mit nicht ganz unzweideutigen Ausdriicken, die Beziehung axif einen 
Inhalt, die Richtung auf ein Objekt (worunter hier nicht eine Realitat zu verstehen 
ist), oder die immanente Gegenstandlichkeit nennen wiirden. . . Diese intentionale 
Inexistenz ist den psychischen Phanomenen ausschliesslich eigentiimlich . . . Und 
somit konnen wir die psychischen Phanomene definieren, indem wir sagen, sie 
seien solche Phanomene, welche intentional einen Gegenstand in sich enthalten" 
(Psych, vom empir. Standpunkt I 115). Wieder ist hier gesehen und aufs scharfste 
betont, dass Psychisches nicht erst an sich, als isoliertes "Datum", besteht, um 
erst nachtraglich in Beziehungen einzutreten, sondern dass die Beziehung bereits 
zu seiner reinen Wesensbestimmtheit gehort. Es ist rrax indem es, in eben diesem 
Sein, gewissermassen iiber sich hinaus ist, indem es auf ein anderes geht" . Cf. 
1. c. 141 sqq., 365 sqq. 


a referent of which they have the most dissimilar ideas, if they only 
refer to the same object all the time. "What for one is no more than 
a kodak, develops for the perception of another into a 'reflex, exten- 
sion, swing-front, focal-plane, anastigmatic hand-camera de luxe' " 
(Spearman 257). The word cannot reasonably be said to have the 
same meaning for the two persons, yet they are able to discuss 
camera because they refer to the same thing by the same word, 
although they do it in different ways. 

The objective reference is thus the constant element in meanings, 
however the subjective apprehension (see next paragraph) may vary. 
Those theorists who have operated in their analysis with word and 
meaning only, neglecting the referent, have, in seeking for the constant 
element in the meanings of a word, been compelled to assume an ab- 
straction variously termed Begrijfskern, Kernbedeutung, begrifflicher In- 
halt, etc., which was supposed to recur unchanged in the meaning of 
a word in all its occasional uses. Nobody has been able to define this 
Begrifjskern in a satisfactory manner: it has remained a phantom of 
logical abstraction. The theory of meaning adopted here takes the 
empirical view and bases meaning on the referent. The constant ele- 
ment in the meaning of the word camera, whenever used, is the fact 
that the word is referred to one or more of the objects belonging to 
the category of 'cameras'. That category is an empirical fact, the 
existence of which a philologist can simply take for granted. It is a 
problem for epistemology to analyse its origin, formation, and charac- 
teristics (cf. 4.16 note). 

It is evident that there cannot normally enter into the meaning of 
the word camera any element of thought that is not founded on an 
actual characteristic of the referent, the camera. But it is not neces- 
sary that all such characteristics shovild be represented in the meaning 
(see 4.24 on specialized meaning). It is further evident that when the 
word camera is used of different cameras, the meaning changes in cor- 
relation to the change of referent. The sentence there flies a bird has not 
the same meaning when used of a fluttering sparrow, of a swallow, 
an eagle, and so on, in a variety of circumstances. Although the words 
remain unaltered, the meaning changes with the change of referents. 

I conclude that the meaning of a word is determined by the 
characteristics of the referent, which is thus a necessary factor in 
the differentia specifica of meaning. 


The symbolic function of a word is conditioned by the reference of 
the word to that which it denotes, and by the fact that the meaning 
is representative of the characteristics of the referent. 

3.24. The Subjective Apprehension. It is easy to see that the ob- 
jective reference is not sufficient to determine the meaning of a word. 
The perception of a fluttering sparrow may lead to quite different utter- 
ances on the part of different persons: there flies a bird, there is some- 
thing moving, how frightened it is. The referent has more than one 
characteristic, verschiedene Merkmale (Wundt), mehrere intelligible 
Teile (Gomperz II. i. 63. See Husserl II. i. 47, 418, Marty Unt. 437, 
Messer Arch. 148, Schwarz 155 sq., Cassirer, Phil. I 44, 252 — 253; 
Gomperz, Sinn 83; Paulhan, Revue Phil. 104.40). 

Meaning may vary although both word and referent remain imchanged 
"Die Vorstellung die ich von Gronlands Eiswiisten habe, ist sicherHch 
eine andere als diejenige, die Nansen von ihnen hat", says Husserl (II. i. 
418). A speech may make one listener laugh and another weep (Otis 
410). The word fight is apprehended differently by a pugnacious and a 
peaceable man (Gomperz II. i. 223, 225, 231). The same word will of ten 
be used of the same referent whatever emotion or conation is to be 
expressed (Buhler Satz 6; he calls this das Prinzip der Dingkonstanz; 
Nyrop, IV 22, calls it the individuahty of meaning, and gives French 
instances; see also Allers 15). "Keiner deiik± bei dem Worte gerade das, 
was der Andere, und die noch so kleine Verschiedenheit zittert, wenn 
man die Sprache mit dem bewegHchsten aUer Elemente vergleichen 
will, durch die ganze Sprache fort" (Htunboldt § 65. Cf. Marty, Unt. 
436 note, and the instance of the kodak in the preceding paragraph). 

Such variations are explained by the circumstance that the referent 
generally has more than one attribute or characteristic {Merkmale, in- 
telligible Teile). In different contexts different attributes of the re- 
ferent will be relevant to the situation and the speaker's purpose, and 
therefore different items will be predominant in the speaker's attention 
(See 4.24 on speciaUzed meaning). Further, as in the instance of the 
kodak, different persons will be acquainted with a greater or smaller 
number of those characteristics, a circumstance that greatly influences 
their apprehension of the referent. Meaning is essentially personal. 
What a word means depends also on who uses it, when, where, why, in 
what circumstances, with what aim, with what success. "We should 
therefore not be content with the conventional meaning of a word, 


but continue the analysis with the meaning of the man who uses it" 
{Schiller 385 sqq. Cf. Wundt II 505 — 506, Schwarz 155 — 156). 

It is evident from these arguments that the subjective apprehension is 
an indispensable determining factor of meanings. In fact, the subjective 
apprehension of the referent, what the subject thinks and feels about the 
referent, is that portion of the simultaneous mental content in which we 
have to seek the meaning of the word by which the referent is denoted. 

The expressive function of the word is conditioned by the fact that one 
determinant of meaning is the subject's apprehension of the referent.^) 

3.25. The Traditional Range. If two men are looking at, and speak- 
ing of, a boat, they may be of opinion that it is a fine boat, or perhaps 
that it is a very bad boat. Their opinion is mental content referring to 
the boat (the referent), and it is part of their subjective apprehension 
of the boat, but it does not belong to the meaning of the word boat, 
which is a purely cognitive appellation. If they want to express 
their total apprehension of the boat they must have recourse to other 
words, in due syntactical order. 

On the other hand, if we are speaking of a person's scrawl or scribble, or 
cackle or chatter, we are not only naming his hand-writing or conversa- 
tion, but we are also expressing an opinion of them. 

Such differences, as well as others of a similar type, are due to the 
origin and history of the individual words, which, in the course of linguis- 
tic development, have led to a consensus in the speaking community 
concerning the referents that a word can normally be used to denote, 
and the mental content (the subjective apprehension) that it can nor- 
mally be used to express (see 2. 11 on the language system). Each word 
has a traditional sphere of application; I shall call it the range of the 
word. The term has a purely quantitative import, and it is not equi- 
valent to the meaning of an isolated word, or to the general concept 
that the word may denote (see on these points 4.22 below). The range 
of the word is either the totality of referents that may be denoted by it, 
the referential range, or the totality of meanings that it can express, 
the semantic range. 

The instance of the boat shows that the objective reference and the 

^) The statement that the referent may remain unchanged although words 
and content change, may require some qualifications. Otto (41) accepts this 
statement, quoting Husserl and Marty. See also Ward 32 and 60. But cf. AUers 
16; Griinbaum, Arch. 36, 438, 441: 5. 


subjective apprehension are not sufficient to determine definitely the 
meaning of a word, distinguishing it from all simultaneous mental con- 
tent that does not belong to the meaning. It is necessary to introduce 
the traditional range as a third, and final, determining factor. 

Any mental content simultaneously present in mind, but not falling 
within the traditional range of the word as apprehended by the user, 
does not belong to its meaning, however closely it may be associated 
with the word or the referent. It is necessary to add, "as apprehended 
by the user", because it is very seldom that the total range of a word 
is present in mind (cf. 4.22). This is equivalent to saying that to 
the meaning of a word belong only those elements of mental content 
that actually are expressed by it. 

The range of a word is normally stable, but it may be affected by 
various factors, so that items are added or deducted. An addition to 
the range is a sense-change. 

The essential importance of the traditional range and its relative 
stabiHty is evident from the fact that it is the indispensable condition 
for one of the three functions of the word: the communicative function 
(see 2.12 above). It is obviously necessary for speaker and hearer to 
have approximately the same notion of what a word may denote or 
express, if they are to have any chance of mutual understanding. 

The existence of a traditional range for all words in a language makes 
the language a normative system for the meanings, and it is this fact 
that makes semasiology a branch of linguistics; it would otherwise fall 
under the heading of psychology', since the two other determinants, the 
objective reference and the subjective apprehension, are purely psycho- 
logical in nature (cf. Ahlmann 73, Cassirer, Phil. I 22). 

The totality of traditional ranges in a language is an element of the 
language system (2.1 1). I have already pointed out that each individual 
belonging to a speaking community must conform to the prevailing 
language system if he wishes to be understood. The correlation be- 
tween the transsubjective normative system, and the individual real- 
izations of it, is not complete, but reaches a comparatively high 
standard (Ahlmann 87). 

3.26. Summary. A Fourth Factor? We have arrived at the conclu- 
sion that there are three determinants of meaning. 

We have found, in the first place, that the objective reference is an 
indispensable element in any meaning, and that the quaUtative charac- 


teristics of meaning are conditioned by the actual characteristics of the 
referent which the word is employed to denote. This factor conditions 
the symbolic function of the word. 

We have found, secondly, that the meaning of a word is determined 
also by the subject's apprehension of the referent that the word is em- 
ployed to denote, that is to say, the subject's thoughts and feeHngs with 
regard to the referent. This factor conditions the expressive function 
of the word. 

We have found, thirdly, that the traditional range of a word serves 
to discriminate its meaning from concomitant elements of mental con- 
tent, or mental context. The meaning of a word in speech normally 
lies within its range. This factor conditions the communicative func- 
tion of the word.^) 

I stated above that the purposive function is a characteristic of the 
utterance, perhaps of the sentence, but not of the word (2.13). We 
have here a possibility of testing this view. If the purposive function 
was a characteristic of the word, it is reasonable to assume that it 
would, like the other three functions, be directly correlated to a deter- 
minant of meaning. The three determinants of meaning described in 
the preceding paragraphs would not then form an adequate differentia 
specif ica; we should have to add the relation of meaning to listener as a 
fourth determining factor. As fas as I can see, this is not the case. It 
is true that the thought of the hearer has a great influence on what is 
said and how it is said (cf. 4.155 and 5.52). The audience influences 
the speaker's choice of words, his tone, tempo, and gestures, in 
short, the way in which speech, as an instrument for the purposive 
function, is handled. The attitude of the speaker towards referent 
and hearer is the basis of this function. It must unavoidably influence 
the meaning of the utterance as a whole, which the hearer gathers, 
not only from the meaning of the single words (see 6.33) but from all con- 

^) It is not uninteresting to quote here Humboldt's statement, "In dem Be- 
griffe liegt ein Dreifaches, der Uindruck des Gegenstandes, die Art der Aufnahme 
desselben im Subjekt, die Wirkung des Worts, als Sprachlaut" (Versch. § 61). Cf. 
also Delacroix' statement that language has its roots in three factors: the condi- 
tions of communal life, the laws governing the expression of emotions, and the 
structure of human intelligence. They are fundamental and irreducible (Dela- 
croix 47). If we substitute "mental content" or perhaps "thoughts and feelings" 
for "emotions", the three factors given by Delacroix correspond to those enu- 
merated above. 


comitant circumstances. It is in this combination that the purposive 
function is involved. Any influence that it might exert on the word- 
meaning would thus be primarily a matter of subjective attitude and 
that is a factor already included in the arguments. 

3.27. Definition. I conclude that the meaning of a word, in any 
individual case of actual use in speech, is completely determined by its 
relation to the three factors, word, referent and subject. Each of these 
three factors conditions one of the three functions of words. No other 
determinant is necessary for the definition of meaning, but all three 
must be embodied in the definition. I propose the following formula- 

The meaning of a word — in actual speech — is identical with those 
elements of the user's {speaker's or hearer's) subjective ap-hrehension of 
the referent denoted hy the word, which he apprehends as expressed by it. 

The definition is appUcable to prefixes, suffixes, and stem syllables, 
in so far as these carry a distinctive element of meaning. 

It is not applicable to speech as a whole (utterance,) owing to the 
purposive function not being included. 

It is further not applicable to signals or substitute signs (2.22). 

It will be the task of the following chapter to elaborate this definition 
by filling in the outUnes and discussing the individual modes and va- 
riations in which meaning appears. 


4.1. Psychic Elements and Categories of Meaning. 

4.11. Psychological and Logical Points of View. The meaning of 
a word, as defined in the preceding chapter, may be analysed from 
various points of view. 

We may study it with regard to the psychic nature of the mental 
content constituting it. This may conveniently be called the psycho- 
logical point of view. 

We may study the meaning with regard to its notional content, and 
to the different elements that constitute this content. For instance, 
we say that swift means "moving, or capable of moving, at high speed". 
For lack of a better term I shall call this the logical point of view. (See 
especially Gomperz II. 1.2 and 6 sqq., Moore, Mono. 184 — 185). 

The following paragraphs will serve to show what belongs to one 
or the other of these aspects of meaning, which it is very important 
not to confuse. I shall first give an account of the psychic elements 
of meaning. 

4.12. Cognitive and Emotive Elements. The first distinction to be 
made here is that between cognitive and emotive elements. "In actual 
discussion terms are used at least as much for the sake of their suasory 
and emotive effects as for their strictly symbolic value. . . It is often,, 
indeed, impossible to decide whether a particular use of symbols is 
primarily symbolic or emotive. . . Most terms of abuse and endearment 
raise this problem, which, as a rule, it is, fortunately, not important to 
settle. The distinction which is important is that between utterances 
in which the symbolic function is subordinate to the emotive act and 
those of which the reverse is true. In the first case, however precise 
and however elaborate the references communicated may be, they can 
be seen to be present in an essentially instrumental capacity, as means 
to emotive effects. In the second case, however strong the emotive 
effects, these can be seen to be by-products not essentially involved in 


the speech transaction. The peculiarity of scientific statement, that 
recent new development of linguistic activity, is the open avowal of 
its restriction to the symbolic function" (Ogden-Richards 226 — 227). 

The term 'symbolic' corresponds to what I have called 'cognitive'. 
It is true that any cognitive element of meaning must be, by definition, 
equivalent to an element of the language-user's apprehension of the 
referent, and in so far is representative or symbolic of the latter. But 
I prefer to use the term symbolic with regard to the word, the verbal 
symbol, and to speak of cognitive elements in the meaning.^) 

4.13. Thought and Imagery. With regard to the cognitive elements 
of meaning and of mental content in general, it was for long an accepted 
opinion that they were composed of sensations and their reproductions, 
the images. The images might be comparatively vivid and clear, or 
they might be faded and highly transient. 

This view was at no period entirely unopposed, and about thirty 
years ago psychologists began to pay particular attention to the 
problem, almost simultaneously in Germany (the Wiirzburg School, 
O. Kiilpe and his pupils), in France (A. Binet) and in the United States 
(Woodworth, see Psych. Rev. 22, 1915). These scholars asserted the 
existence of "imageless thought" (des pensees sans images, unanschau- 
liches Denken) as a third kind of mental content, essentially distinct 
from images, although like them of a cognitive nature. Imageless 
thought is the main constituent of meanings and of the higher thought 
processes in general. 

On the other hand, Titchener, G. E. Miiller, and other scholars, assert 
that when observers report after introspection that they found nothing 

^) "In general, we may say of any act that is being performed that it is a cer- 
tain specific sort of act. It has an end; the animal is engaged in doing something. 
The end may be more or less remote, and the behavior leading up to it more or 
less involved, but it is generally pertinent to ask the question. What is the animal 
doing? In addition to this specific character, behavior has a particular form. 
The animal is not only doing something, but he is doing it in a particular way. 
In so far as behavior can be characterized as even crudely intelligent, it involves 
a certain adaptation of means to an end. These two aspects of behavior form 
the basis for the distinction between affection on the one hand, and cognition on 
the other. It is the affective properties of situations and things which determine 
the ends of action, and in the most general sense of the term it is the cognized 
properties and relations which determine the particular form which the action 
takes, the means by which the end is achieved. Affection and cognition then, in 
the sense in which the terms are here used, are aspects of all behavior" (Laguna 190). 


in their mind except an awareness, a knowledge, of some kind, this is 
due to their inability to distinguish the visual, auditory or motor imagery, 
which, although probably very fugitive, was really present to them.^) 

It is evident that when we are speaking of concrete things, especi- 
ally if they are familiar to us, visual images may, and often do, accom- 
pany speech. But even this is no proof that the meaning of the words 
is constituted by the images (see next paragraph). And with regard 
to abstract words and phrases, how could the meaning of philosophy , 
nothing hut, nevertheless, although, differential calculus, etc. be carried 
by imagery of any kind whatever? (James II 52, Husserl II. i. 62 — 
63). Explanations in terms of muscular sensation or innervations, 
whether in the articulatory organs or in other parts of the body (cf. 
3.14), seem to many scholars entirely unsatisfactory. 

To me, the "imageless" theory appears attractive, but it is for- 
tunately not necessary for my purposes to go into the matter in detail. 
I shall speak of thoughts and images, denoting with the latter term 
undoubted images, mainly visual; and the reader may, as he pleases, 
interpret my term 'thoughts' as signifying either imageless thoughts, 
or imagery of a peculiar type. There are evidently great individual 
differences with regard to imagery, which may account in part for the 
conflicting theories.^) 

1) Cf. for instance the following quotation from R. H. Wheeler, Development 
233: "In every stage kinaesthesis is the core of the phenomenon which we call 
meaning ... In the writer's case, at least, meaning has not been described until 
kinaesthesis is taken into account. It is quite probable that pure meanings, so- 
called, are in reality masses of diffuse muscular sensation which the reagent has 
not succeeded in recognizing and describing". But it is very difficult to get a 
clear idea of what Wheeler really means by 'meaning'. 

2) I may refer to the following works, among a much greater number: Spear- 
man 175 sqq., (with very useful discussions), Moore, Mono.ioi sqq., 154, 234 sqq., 
254, Moore, Psych. Rev. 22 and 24, Stout I 83, II 194, 211, Titchener, Lectures 
9 sqq., R. M. Ogden, Am. J. of Ps. 34, Kantor, Am. J. of Ps.32, 231 sqq., Ogden- 
Richards 150 sqq., Laguna 177 sqq., 198, Pillsbury-Meader 100 sqq., 152 sqq., 
162, 183, Binet, passim, Bourdon 189 sqq., Husserl II. 1.65 4), 61 sqq., Gomperz 
II. 1. 170 sqq., 180 sqq., 241, G. E. Miiller III 520 sqq., Watt, Arch. 4, 433, Messer, 
Empf.u.Denken 92 sqq., and Arch. 85, 175 sqq., Biihler, Ber.io6, and Arch. 9 298 
sqq., 317, 350 sqq., 361, Marbe, Beitr. 493 sqq., Lindworsky 91 sqq., Stahlin, 
Met. 38, Allers 11 sqq., Poppelreuter 325 sqq., Meyer, Stilgesetz, passim, Pick, 
Sprachstor. 183, MiUler-Freienfels, Psych, d. Kunst I 160 sqq.; WillwoU 64 sqq. 
— Some of these writers take up an intermediate position between the two ex- 
tremes briefly indicated above. 


4.14. Images. 

4.141. Useful, Inadequate, and Irrelevant Images. The part played 
in our mental process, particularly in meaning, by auditory and motor 
images seems at present to be so vaguely known or at least so disputed, 
that it is impossible to make any definite statements on the subject. 
I shall therefore leave them aside, and confine myself to a discussion 
of visual images, by which term I mean images plainly showing their 
character as such. They have been the subject of eager study, and 
are comparatively well known. 

Visual imagery in many cases clearly serves as an aid to understand- 
ing, by providing illustrations and making the realization of the re- 
ferent more vivid and tangible. In one of Stahhn's experiments, an 
observer reported that the word holy was accompanied by the image 
of a temple. For another observer, the idea of friendship was illustrated 
by an image of Abraham, because a sermon on Abraham as God's 
friend had once made a deep impression on the observer (StahHn, Unt. 
166 sqq.). Other instances of useful imagery are given by Selz (II 120; 
cf. Willwoll 72 sqq.). The genus proximum for star was found by one 
observer with the help of a visual image of a star against a dark back- 
ground; the quality of brilliancy noted in the image led to the correct 
solution: luminous body. (See 6.22 on imagery as mediating the 
comprehension of ' words) . 

As an instance of inadequate imagery we may quote a case from 
Binet (90) . He was speaking to one of his observers (Marguerite) about 
their dog, which had recently died. He said: "what a pity that all ani- 
mals without exception have to die!" Then he asked, "Quelles 
images?" Marguerite was a little surprised, and at first said that she 
had had no images. After some reflection, however, she remembered 
having had a faint image of a black, shrunken insect. It is evident 
that this is a very inadequate illustration of the speech made (Other 
instances are given by Btihler, Ber. no, Allers i, Selz II 124, 163). 

Images, both of words and things, may be altogether discrepant. 
(Spearman 183, who also quotes Binet). Other writers assert that 
images are never irrelevant. (See Comstock and Tolman, and cf. 
Willwoll 68). 

Disturbing imagery is instanced in the following report from Stahlin. 
The experimenter read a text containing the passage: "Gleichwie aus 
vielen Kohlen ein grosses Feuer und daraus eine richtige Flamme ent- 

Goteb. Ho^sk. Arsskv. XXXVIII: i 


steht, die in die Hohe, weit iiber die Kohlen hinflackert, also soil der 
Mensch sein Gemtit durch alle Gedanken, Bilder und Werken seiner 
untersten und obersten Krafte hindurchdringen lassen und in die Hohe 
weit iiber sein eigenes Konnen und Wirken sich emporschwingen in 
jene unbegreifliche Hohe der iiberwesentlichen Gottheit selbst." Sev- 
eral observers report that the image of coal and fire occupied them so 
much that they had difficulty in grasping the import of the passage. 
One of them stated that at first he saw only coal, but no fire, then a 
coal-yard in K. Street, smoking but not burning, then an image of 
the fire in the Meiningen Theatre, and finally of a lark rising towards 
the sky! (Stahhn, Exp. Unt. 129, 136, 149; cf. Biihler, Ber. 123 sqq. 
Willwoll 70 sqq. and Delacroix 385 sqq.) 

Stahlin (1. c. 161 sqq.) arrives at the conclusion that in many cases 
imagery is necessary, or at least valuable, for comprehension. Accord- 
ing to Betts (94 sqq.) images occur especially when the thought pro- 
cess encounters an obstacle, and images are resorted to as a help for 
overcoming it. Effective and successful thinking, according to Betts, 
is with most persons accompanied only by a minimum of images. Bur- 
kamp (Kaus. 262) also states that images are fewest when the mental 
work is most energetic. This is due to the fact that images require a 
certain time to develop, which time is not always available. 

The essential difference between useful imagery and disturbing 
imagery seems to be the circumstance pointed out by Stahlin and 
others, that images which are determined by the import of a speech and 
stay within its sphere of meaning, serve to illustrate and elucidate it. 
(Cf. Selz II. 184 — 185, Gomperz II. i. 241). But disturbing images 
are nearly always irrelevant to the trend of thought expressed in a 
speech. Images of the former kind may, however, also have a disturb- 
ing effect if they become too obtrusive, and thus attract too much 
attention and give a disproportionate prominence to some detail. 

I conclude that, as Delacroix states, the image is fragmentary, 
arbitrary, and accidental. "Elle symbolise le travail mental, plutot 
qu'elle ne I'exprime. Elle n'en represente jamais que des fragments. 
Elle survient surtout quand I'operation est effectuee ou pres de s'ef- 
fectuer; ou bien quand il est necessaire de la jalonner de points de 
repere, ou quand une difficulte survient. Elle abonde aussi quand la 
pensee se detend, se laisse aller" (Delacroix 385). Images are con- 
ditioned by the thought process and presuppose comprehension. An 


image receives meaning only through being interpreted and related 
to other images, words, or notions (1. c. 386). 

4.142. Images as Signs, Meaning, or Context. Head (Aphasia 
I 520) has made an interesting distinction between two kinds of im- 
agery which behave differently in disorders of speech. "In persons with 
a strong visual memory all the processes of thinking are accompanied 
by and at times essentially composed of more or less vivid and detailed 
imagery. If I think of a horse, it is not the word in any form which 
springs into my mind, but a picture of a horse. This image assumes a 
familiar general character, which usually represents a horse to me; 
it is in reality a nominal symbol or a visual noun. If it has been 
aroused by something I have heard or read, the figure is suitably varied 
in colour, form or posture in accordance with the descriptive details, 
and in this way reproduces adjectival meaning. Such images stand 
in the place of words and as such tend to be affected in aphasia. 

On the other hand, during spontaneous thinking visual images may 
appear in a sequence suggested by association, or corresponding to 
the order in which the objects were originally perceived. Such images 
form perceptual data, which may remain unaffected in disorders of 
speech .... Even a vivid and accurate series of visual images is insuf- 
ficient alone for constructive and logical thought" (cf. 5. 21). 

Other investigators agree that images alone cannot carry a sequence 
of logical thought, but it seems that they can function in some way as 
signs. Ivike a verbal sign, an image may carry a certain meaning (Pills- 
bury-Meader 157, 182). Moore concludes from his own experiments 
that an image has no meaning in itself but must be interpreted, which 
can be done only when sufficient data are available (Moore, Psych. 
Rev. 22, 177 sqq.) "The interpretation of the image is a knowing. It 
is something which follows the awareness of the image, just as under- 
standing follows the sensations involved in perception" (1. c). Selz 
asserts emphatically that images are evoked by the purposive thought- 
processes tending towards the solution of the task in hand (Selz II 
184 — 185. Also Ogden-Richards 150). This agrees with Moore's con- 
clusion that meaning is prior to imagery. And it seems reasonable 
that, in the instance quoted from Selz, the word star could not evoke 
the image of a star if we had not first comprehended the word, that 
is to say, if we had not become aware of its meaning (cf. 6.22). 

I have no evidence from other sources to confirm Head's statement 

52 GUSTAF STERX 4-142 

that images can function as signs independently of words, but it does 
not seem unreasonable to assume that they can. In general, they 
must be assumed to appear together with words, as supplementary 
signs, a function that is amply corroborated by normal psychology. 
The image "either does not explicate the concept at all, or it does so 
in an utterly inadequate way. It may indeed fulfil a useful function 
in the thought-process, but this function belongs to it as a supplement 
of the word, not as an explication of the meaning. It is a component 
part of the sign rather than a presentation of the object signified. If, 
in reading a treatise on political economy, the word wealth calls up 
in my mind the vague picture of a bale of goods, this picture is, to 
say the least, a hopelessly inadequate explication of the concept of 
wealth. But it may co-operate with the word in fixing and detaining 
the mental system associated with this word. The mental imagery 
that clusters round a word, and supports it in its function, constitutes 
what has been called the 'inner speech form' " (Stout II 211, and 
similarly I 85).!) 

The researches on the process of understanding, quoted in 6.22 
below, show that meaning may be represented by imagery, and that, 
just as in the case of a word, sign and meaning may fuse. The image 
seems to be the referent itself. In these cases, too, the word is present. 

The usefulness of images as signs is limited by their inability to per- 
form the expressive and communicative functions: they remain within 
the confines of the individual mind.^) As soon as the word appears 
on the scene, which naturally happens in most cases, the image is 
therefore liable to be relegated to the position of a supplement, more 

^) Stout, in the passage quoted, speaks of imagery clustering round the word, 
fixing and detaining the mental system associated with the word, and so on. I 
think we should perhaps substitute the referent, since a word may have more 
than one meaning. For a speaker, the awareness of the referent is at least not 
posterior to the word, and an image may be prior to it. For the hearer, as point- 
ed out above, the word must normally be understood in order to call up an image, 
that is to say, the hearer must know what referent it denotes. 

") "The hnage . . . just because it is bound up with a particular response of 
bodily adjustment, can control only a direct sort of behavior, i.e., behavior which 
is determined by the relations of things to the actor. Behavior which is indirect, 
in that it is determined by the objective relation of things or of persons to each 
other, and which issues in the production of an objective state of affairs of mediate 
utility or interest to the actor, must be controlled by something other than the 
image" (I^aguna 298). 


or less useful, and it is thus evident that images are not on a par 
with verbal signs.^) 

When images are not signs, are they to be reckoned as elements 
of meaning or of context? If we take the image of a horse, 
illustrating the thought of a horse, it undoubtedly falls within the 
first two criteria of meaning: it refers to the referent, and it belongs 
to the speaker's subjective apprehension of it. But it is excluded by 
the third criterion, for it is not expressed by the word. Normally at 
least, the speaker does not indicate in any way what images may 
have accompanied his train of thought, and it is not probable that 
the hearer will spontaneously evoke the same images. If we accept 
the definition of verbal meaning given in the preceding chapter, we can 
scarcely avoid the conclusion that visual images must belong to con- 
text. This agrees with Stout's position. When an image of Abraham 
illustrates the meaning of friendship, such an image cannot belong to 
the meaning of the word, although it may be useful to the speaker 
in fixing the notion. Irrelevant imagery is not even useful. 

The conclusion is that images seem to belong to mental context 
(on context, see esp. 6.32). They may sometimes function as supple- 
mentary signs. 2) 

1) Cassirer also emphasizes the superiority of the word as symbol: "Das Spracli- 
wort unterscheidet sich eben darin vom sinnlichen Anschauungsbild, dass es 
sozusagen mit keiner eigenen sinnlichen Materie mehr belastet ist. Betrachtet 
man es seinem blossen sinnlichen Bestand nach, so erscheint es als ein Verschwe- 
bendes und Unbestimmtes: es ist ein Spiel von jedem Hauch der Luft. Aber 
gerade dies Ungreifbare und Fliichtige an ihm begriindet zugleich — vom Stand- 
punkt der reinen Darstellungsfunktion gesehen — seine uberlegenheit iiber die 
unmittelbar-sinnlichen Inhalte. Denn das Wort besitzt sozusagen keine fiir sich 
bestehende selbstandige 'Masse' mehr, mit der es der Energie des beziehentlichen 
Denkens Widerstand leisten konnte. Es ist fiir jegliche Form offen, die der Ge- 
danke ihm aufpragen will" (Cassirer, Phil.III 383 — 384). 

2) Marty's position with regard to imagery is not clear. He defines meaning 
as "dasjenige psychische Phanomen, welches der sprachliche Ausdruck ihi Horer 
wachzurufen bestimmt ist" (Funke, Sprachform 20 — 21). As far as I have no- 
ticed, Marty discusses imagery only with regard to transferred and metaphorical 
expressions, where he calls it figurliche innere Sprachform, and only with regard 
to the hearer, in accordance with his usual practice and with the definition just 
quoted. Funke (I.e. 26) gives as an instance the word lion, applied to a heroic 
person. In the hearer, the word at first evokes the image (Vorsiellung, which I 
suppose is to be understood as visual image), and through the mediation of the 
image of a lion we get to the Vorstellung (image?) of a man of certain qualities 

54 GUSTAF STERN 4- 15 1 

4.15. Emotional and Volitional Elements of Meaning. 

4.151. Preliminary Remarks. Human speech in its earlier phases 
was no doubt primarily emotive; that is the case with children's speech. 
The cognitive element was present from the beginning, and in some 
styles at least, it now plays the most important part. But nevertheless 
there is some sort of emotive element in all speech; if a thing were 
quite indifferent to me I would not say it (Delacroix 374).^) 

The emotional elements of meaning correspond to the subjective 
attitude taken up by a speaker or hearer towards the referent and 
the interlocutor (cf. Ahlmann yj). The attitude may vary indefi- 
nitely: the same referent may be spoken of with doubt, concession, 
desire, questioning, pleasure or the reverse, contempt, irony, derision, 
confidence, and so on. How these var)dng attitudes are expressed and 
how they influence the meanings of the words used is an immense and 
complex problem that has as yet scarcely been touched by scientific 
investigation. It is therefore impossible to give anything like an ade- 
quate account of the matter. I shall have to confine myself to an 
account of some distinctions which have proved of essential impor- 
tance in the analysis of meanings and sense-changes (cf. Biihler, 
Ber. 112 sqq. and especially Ahlmann 51 sqq.). 

4.152. Methods of Expressing Emotive Elements. When we turn 
to emotive elements in the meanings of words — the statements of the 
preceding paragraph apply to speech — we have to keep in mind 
not only the distinction between word and speech, but also that 

The image (of the lion) is not an element of the meaning of the word lion, but a 
means of evoking the meaning (in this case the thought of a heroic man), according 
to the laws of association (I.e. 27). Marty's analysis of the process of understand- 
ing will be criticized below (i 1.59). It will be found that in cases like the one quoted 
an image of the lion is probably not a necessary, or even a normal, phase of the 
process; on the contrary, an image representing the original meaning of a meta- 
phorical word would in most cases probably be a disturbing factor. 

1) According to Shand, an emotion always involves "(i) a cognitive attitude, 
in the sense of a perception or a thought, (2) a conative attitude, in the sense 
of an impulse and an end, and (3) a feeling-attitude, of a peculiar kind which 
we cannot fully analyse" (Shand 64). "An emotion is an attitude of the mind — 
a perception or a thought, not merely sensation . . . All moods of emotion arise 
at first without a defined object, but there is an inherent tendency in them to 
search for one, because an object is necessary to organize and direct their im- 
pulses" (I.e. 272. Cf. Ward 42 sqq., 275 sqq.; K. O. Erdmann 103, Sechehaye, 
Progr. 89, Moore, Mono. 98, Stout I 272, Pick, Sprachstor. 117 sqq. 


between symbols and signals (2.22). I am concerned primarily with 
the former. 

Verbal symbols may serve to express emotion in two ways, (i) The 
emotion (the subjective attitude) may be the referent of the word, as 
in / am very much annoyed at your coming here, where annoyed denotes 
the emotion, of which the hearer is thus directly informed. (2) Some 
words possess an emotive colouring as a permanent element of their 
meaning; instances are relatively numerous; compare horse and steed, 
-poor young man and wretched wight, house and hovel; bright, gaudy, gor- 
geous and flashy, where different attitudes to identical referents are 
expressed. Fr. jrapper denotes the action of striking, hattre implies 
that the action is brutal or dishonourable; animaux are a category of 
living beings, betes are of interest to us: pauvres betes! lyatin niger is 
black colour, in contrast to albus; ater is black and gloomy, Uke spilt 
blood or the dreary night. The Nile, with its fruitful slime, is niger, 
but the infernal Styx is ater (Marouzeau 561 sqq. with instances also 
of emotive expression through word-order or accentuation; cf. Bally I 
170 sqq.). 

In both these cases, the emotive elements belong to the meanings 
of the words (cf. a different analysis in Ogden- Richards 356 sqq.). In 
the first case, where the emotion is the referent, the apprehension of 
the emotion must be a central element of meaning (cf. 4.16). In the 
second case, the emotive elements may be central or peripheral. 

Signals (as described above 2.22) may be verbal or non-verbal, and 
are probably our most important instruments for expressing and com- 
municating emotions. The same emotion that is expressed in the sen- 
tence / am very much annoyed at your coming here, may also be ade- 
quately expressed by the two words You here! spoken with the appro- 
priate tone, and perhaps accompanied by appropriate gestures. The 
same words, with different tone and gestures, may express fear, plea- 
sure, surprise, and so forth. The signal system runs parallel with the 
symbols and is able to turn identical symbols into expressions for very 
different emotions. (Cf. Bally I 7, and Ahlmann 76).^) 

1) Husserl excludes from Ausdriicke (approximately equivalent to what I have 
termed symbolic signs) "das Mienenspiel imd die Geste, mit denen wir unser 
Reden unwillkiirlich und jedenfalls nicht in mitteilender Absicht begleiten, oder 
in denen, auch ohne mitwirkende Rede, der Seelenzustand einer Person zu einem 
fiir ihre Umgebung verstandlichen "Ausdrucke" kommt. Solche Ausserungen 

56 GUSTAF STERN 4-152 

The non-verbal signals belong to tbe spoken language. "In written 
language many of the most obvious signs for these attitudes are ne- 
cessarily lost. Manner and tone of voice have to be replaced by the 
various devices, conventional formulae, exaggerations, under-statements, 
figures of speech, underlining, and the rest, familiar in the technique 
of letter-writing. Word-order is plainly of especial importance in this 
connection" (Ogden- Richards 357). 

From the point of view of verbal meanings — my point of view in 
the present study — signals and their meanings belong to context. 

The third instrument of emotive expression is phrase meaning (cf. 
4.27). The interrelation of word meaning and phrase meaning is very 
httle known, but it seems probable that a very large proportion of 
emotive expression through symboHc means belongs to phrase 
meaning. Changed word-order, aposiopesis, ellipsis, redundancy, 
are all of them regular methods for the purpose. 

The distinctions which I am seeking to establish here are especially 
difficult owing to the tendency of emotion to permeate all simultaneous 
mental content.^) Nevertheless, I believe that the distinction is ne- 
cessary, and that it is useless to discuss whether the emotive elements 
belong to meaning or not, without making clear whether we are speaking 
of the meanings of words, of phrases, or of signals. In accordance with 
my general programme, I confine myself to word meanings.^) 

4.153. Emotion and Communication. We have to note a peculiarit}'- 

sind keine Ausdriicke im Sinne der Reden, sie sind nicht gleich diesen im Be- 
wusstsein des sich. Aussemden mit deni geausserten Erlebnissen phauomenal 
eins; in ihnen teilt der eine dem anderen nichts mit, es fehlt ihm bei ihrer Aus- 
serung die Intention, irgendwelche "Gedanken" in ausdriicklicher Weise hinzu- 
stellen, sei es fiir andere, sei es auch fiir sich selbst, wofern er mit sich allein ist. 
Kurz, derartige "Ausdriicke" haben eigentlich keine Bedeutung. Daran wird 
nichts geandert dadurch, dass ein zweiter unsere nnwillkiirlichen Ausserungen 
(z.B. die "Ansdrncksbewegungen") zu deuten, und dass er durch sie iiber unsere 
inneren Gedanken und Gemiitsbewegungen mancherlei zu erfahren vermag. Sie 
"bedeuten" ihm etwas, sofern er sie eben deutet; aber auch fiir ihn haben sie keine 
Bedeutungen im pragnanten Sinne sprachlicher Zeichen, sondern bloss im Sinne 
von Anzeichen" (Husserl II.1.31). 

^) Das Gef iihl aber ist jederzeit auf alles unmittelbar bezogen, was gleichzeitig 
oder in erlebter Nachbarschaft damit vorgefunden wird. Krueger, Arch. 65, 114. 

^) Several authors have contended that the emotive elements do not belong to 
meaning: but what kind of meaning do they refer to? See Ahlmann 50 sqq.; 
Noreen, Betr.207, criticized by Ahlmann 55 sqq. 


of the emotive elements of meaning with regard to the communicative 
function. No element of mental content can be directly communicated 
to a hearer: it must be phrased in speech, a totally different medium, 
and be translated back into thoughts. Nevertheless, in so far as speech 
is symbolic of facts, and so expressive of cognitive knowledge, adequate 
formulation leads to a reasonably exact reproduction of the same know- 
ledge in the hearer's mind. The speaker's attitude towards the re- 
ferents is another matter; the hearer is informed of it, say, by verbal 
means, but whether this leads to a reproduction of the same attitude 
in the hearer is not at all certain, and depends not on the ade- 
quacy of the formulation, but on quite other factors. Making use of 
the distinction between proposition and judgement (cf. Husserl II. i. 78) 
we may say that the import of the proposition is communicated, but 
whether the hearer accepts the judgement depends on him, not on the 
speaker. If speaker and hearer are actuated by similar motives in 
respect of the referent, or if the speaker has some influence over the 
hearer, the latter's emotions are likely to be stimulated in sympathy. 
A perceived excitement is to some extent contagious (cf. 6.5 and Ahl- 
mann 77 sq.). On the other hand, the hearer may consider, for instance, 
the speaker's annoyance at his arrival entirely uncalled for, or even 
ridiculous; or he may consider the referent of hovel quite a decent little 
house. Such differences may be due to different factual circumstances. 
Perhaps the speaker is a man who is planning to live in the hovel, 
while his interlocutor is the landlord's agent, intent on making a good 
bargain about the rent. Another not unfamiliar instance is that of 
the pathetic orator who fails to enlist the sympathy of his audience, 
or moves it to derision and laughter. The possibilities are numerous. 
It makes no difference for the hearer whether the verbal symbols through 
which he learns the speaker's attitude function in the first or the 
second of the two ways described at the beginning of the preceding 
paragraph. (Cf. Maier 336). 

4.164. The Sources of Emotive Elements: Permanent Elements. 
Emotive elements in verbal meaning are either permanent or incident- 
al; the importance of this distinction will be evident in the chapter 
on adequation (ch. 14). 

Permanent emotive elements are those that normally belong to the 
semantic range of the word. The subjective attitude of a speaker 
employing such a word is apprehended by a hearer without the help 

58 GUSTAF STERN 4-154 

of further signs; chatter referring to a person's talk is at once under- 
stood as contemptuous. Permanent does not mean eternal, but only 
relatively permanent, common to a group of speakers during a period 
of time; changes may take place, but they are slow and general. Corn 
Laws, Home Rule, Women's Suffrage, are terms that have all of them 
at some time been charged with emotional import for many English 
speakers; at present they are less explosive (cf. 8.4). 

The source of permanent emotive elements may be a referent pos- 
sessing characteristics of more or less constant emotional value to the 
speaking community. Here belong words denoting emotions, affective 
quahties, etc.: fear, anger, furious, indignant, to hate, to love, and so on. 
Another group consists of words like death, resurrection, etc., the re- 
ferents of which normally have an emotional value; German instances 
are given by Meyer (Stilgesetz 160): Glockenklang , Rauch und Qualm, 
Ruhe und Rast. (According to Ahlmann 76 the emotional colouring 
here belongs only to the referent, not to the word. Cf. Britan 49, and 
Oertel 299). A third group consists of words having an emotional tone 
only in certain contexts, i. e., when their referents are apprehended in 
certain aspects. Sometimes the same word is used for the referent in 
any aspect, emotional or not; thus, father may have a purely cognitive, 
"genealogical" meaning, or it may be highly affective; German Weib is 
either poetical or low (Oertel 200. Cf. 4.24 on speciaUzation) . Some- 
times different words are used for the referent in different aspects, 
as house and hovel, and the other instances quoted in 4.152. (Cf. Ahl- 
mann 77, and Meyer, 1. c. 161: Pferd: Klepper: Ross: Rosslein). Sub- 
junctive and optative mood belong here, expressing different attitudes 
towards the referent, that is, the action denoted by the verb. 

The difference in attitude expressed by the two words house and 
hovel is not necessarily a difference between a speaker and a hearer. 
It may be a difference between groups of people. Meillet points out 
that the French verb s'hahiller has not precisely the same shade of 
meaning for men as for women, because the import and character of 
the action denoted are not identical for the two sexes (Meillet, Ling. 
245). Similarly a pugnacious man will regard every chance of a fight 
with pleasure, while the peace-lover will consider the prospect with 
quite different feelings; and the meaning of a statement concerning an 
impending fight will be correspondingly different for the two (cf . Gom- 
perz II. I. 223). Ahlmann (78) points out that some words, as hour- 


geois and capitalist, have a different emotional colouring for different 
classes in the speaking community. 

It has already been noted that the same speaker may react dif- 
ferently to a referent on different occasions, and possibly for that reason 
use different words to denote it. 

When words denote a referent that has a more or less permanent 
emotive colouring, the words themselves become intimately associated 
also with that element of their meaning, and the comprehended word 
consequently possesses a permanent emotive colour. Meyer (Stilgesetz) 
thinks that words with a really permanent emotional colour are com- 
paratively few in number, and quotes as German instances Maid, Leu, 
Minne, minniglich, Kdmmerlein, and other diminutives. (Further in- 
stances, see K. O. Krdmann no sqq.) The speaker chooses words of 
this kind because he is emotionally adjusted; the emotional colouring 
of the words is familiar to the hearer, and he is thus informed of the 
speaker's attitude — which he may share, or not, according to circum- 
stances. It is a task for special historical research to ascertain the 
reasons why a word has an emotional colouring. 

4.155. The Sources of Emotional Elements: Incidental Elements. 
The sources of the incidental emotive elements of meaning should ob- 
viously be sought outside the triangle of subject-referent- word; or, 
since an emotional element corresponds to a subjective attitude, they 
should be sought in the speaker's or hearer's attitude to the interlo- 
cutor or towards outside facts and circumstances. As already stated, 
an emotion has the peculiar power of permeating all simultaneous men- 
tal content, and it is for this reason that the attitude towards the in- 
terlocutor is capable of influencing the apprehension of the referent, 
and consequently of colouring the meaning of a word. A lover's in- 
terest in anything connected with his beloved, even mischievous small 
brothers, is a well-worn theme of comic papers. 

Incidental emotional meaning may further be due to an attitude 
caused by outside factors. A temporary excitement of the speaker 
may lead to the use of words implying an intensity of feeling that is 
very far from being objectively justified, only because the excitement 
permeates the total mental content. Cf. 14.53 below on adequation 
of h5rperbolic statements. 

An important instance of incidental emotional meaning is constantly 
met with in literature. The general topic of a poem or prose compo- 

60 GUSTAF STERN 4-155 

sition will often be of a nature to excite the emotions of the reader 
— and may be conceived by the writer in a similar state of emotion. 
This "Stimmung" permeates the whole text, and imparts an emotional 
colouring to words and phrases that are otherwise totally indifferent 
from an emotional point of view, because the referent is presented in 
such an aspect that it is apprehended emotionally. Almost any poem 
will illustrate the point. The attitude evoked by a text (book) as a 
whole impregnates every part of it. (Cf . Britan's insistence on language 
as an instrument for the excitation of emotions, Britan 49). 

Another instance is the well-known fact that a furious or suspicious 
man will read into the words addressed to him meanings that are far 
from being the speaker's intention, and perhaps logically preposterous 
and unreasonable. 

Incidental emotional elements may be expressed either through 
verbal or through non-verbal means, just like the permanent elements. 

4.156. Volitional Attitudes. Volition is a specific kind of emotional 
attitude. Miiller-Freienfels states that the essential element of voli- 
tion is a selective activity, adjusted to a purpose approved and fixed 
by the subject. (Miiller-Freienfels, Einfluss 384. Cf. Froschels (63): 
Willenshandlungen sind solche die immer einer bestimmten spezifischen 
seelischen Einstellung bediirfen, und mit einer solchen ablaufen). 

Volition, like other emotional attitudes, may be expressed by sym- 
bols (7 want you to come), or by signals (a commanding tone of voice, 
a gesture, etc.). The attitude may be directed towards the referent 
or towards an interlocutor. 

The purpose of volitional expression is not merely to influence the 
state of mind of an interlocutor, but to make him perform definite 
actions — with or against his own will. What was stated above on 
the communicative aspect of emotions is applicable also to volition 
(ci. Maier 19). 

For the rest, the volitional elements of meaning are as little investi- 
gated as other emotional elements, and there is an almost complete 
lack of material for a review of the subject. I must therefore leave it 
with this brief note. (Cf. Paulhan, Rev. Phil. 104, 29.). 

4.16. Central and Peripheral Elements of Meaning. The meaning 
of a word is often complex, and certain of its constituent elements may 
on any given occasion receive a greater share of attention, while other 
elements form a "fringe" (cf. Burkamp 248, with ref. to Westphal, 


Arch. 2i), that is to say, they are less specifically attended to. I shall 
call the former central, the latter peripheral elements. 

It is first necessary to note, that what is central and peripheral in a 
meaning is not at the same time always central and peripheral in mental 
content as a whole. A word may occupy a peripheral position in the 
speaker's or hearer's total mental content, and still have central elements 
of meaning. And, conversely, a word that stands in the focus of atten- 
tion in the mind as a whole, may have peripheral elements of meaning. 

If a builder is speaking of bricks as a possible material for facing a 
building, he is probably thinking mostly of their colour and external 
aspect; if he is speaking of bricks as an alternative material for founda- 
tions, he is thinking of their durability and resistance to high pressure; 
if he is discussing the number of bricks likely to be required for a certain 
construction, he is turning his attention mostly to their size; and if he 
is asking about the number of bricks delivered last week, he will be 
thinking of them as entities, without paying any attention, for the 
moment, to their characteristics (cf. Gompery II. i. 172) 

In this way different elements within the range of a w^ord will on dif- 
ferent occasions occupy a central position because the characteristics 
of the referent which are apprehended through them are relevant to 
the momentary context. On other occasions, the same elements will 
be clearly peripheral, because the corresponding characteristics are of 
no further interest in the actual context. Cf. 4.24 on specialization of 

As already stated (4.12, quotation from Ogden-Richards) emotive 
elements may function as central or as peripheral. It is very common 
for a word to have an emotional tone which makes it suitable for 
use in contexts characterized by a similar tone. For instances, see 
4.152. In discussing hyperboles and adequation we shall come across 
many instances of emotive elements functioning as central, while the 
cognitive elements of meaning have altogether sunk into the back- 

In 4.152 were given instances of words expressing different subjective 
apprehensions of their referents, and therefore applicable only in con- 
nections where such apprehensions are proper and relevant (as wretched 
wight, compared with poor young man. German instances in K. O. 
Erdmann 103 sqq.). We may say that such pairs of expressions have the 
same, or at least approximately the same referential range, but dif- 


ferent semantic ranges (cf. 3.25). According to my definition, the 
emotive elements connected with, for instance, wretched wight, belong 
to the meaning of the expression. I am doubtful whether the charac- 
teristics that make some words suitable in certain styles only, religious, 
poetical, journalistic, vulgar, and so on, are anything but one peculiar 
variety of what I have just been describing. Possibly we ought to 
reckon with special stylistic associations; if so, they would appear to 
fall outside the scope of this work (cf. Ahlmann 83). 

I have stated above (3.16) that the meaning of words comprises only 
a portion of the total mental content present in the mind of a person 
when they are pronounced or comprehended; and I have used the 
term context to denote such elements as are not expressed by the 
words (cf. especially 6.32). The distinction between peripheral elements 
of meaning and context has not always been clearly realized, and re- 
quires some comments. 

One item in context is one's awareness of surrounding circumstances, 
not connected with the topic of speech (cf. 3.22). The awareness of an 
interlocutor is sometimes of great importance for the selection of words 
(5.52). From the hearer's point of view, it makes a great difference if 
a statement is made by a person in whom he has confidence, or by one 
whom he considers as unreliable. Such factors are effective through the 
medium of the subjective apprehension, and they should be considered 
as elements of context, not belonging to meaning although exercising 
an influence upon it. 

We have further the awareness of sphere or direction (Spharenbe- 
wusstsein, cf. 6.22). Harmodios may evoke the thought of his country 
and period, and more specifically may make us think of Aristogeiton, 
perhaps also of Hipparchos, of ancient Athens, of the well-known group 
representing the tyrannicide, and so on (cf. 3.16). All this certainly 
contributes to making the referent of Harmodios stand out more clearly 
in all its implications, by providing it with a fuller background (cf. 6.22 
and 6.36 on mediating items in comprehension). In metaphors, for 
instance, such an associative background is clearly of the greatest 
importance. But it cannot belong to meaning, since it does not fall 
within the range of the word. It should be described as mental context, 
which, through the medium of the subjective apprehension, may exer- 
cise a certain influence on the meaning of the word. 

One point may be noted. It is doubtful whether such elements of 


context are, normally, connected with single words in speech; it seems 
more probable that they rest upon entire utterances, or perhaps 
sentences. The experiments with isolated words cannot always show 
what actually happens in connected speech. (Further details on con- 
text are given in 6.32). i) 

4.17. The Vagueness of Meanings. Jaberg (Herrig's Archiv 136, 
96 sqq.) describes two different reasons for uncertainty concerning the 
meaning of a word. One of them is illustrated by the case of a girl from 
Berne, who was asked, "What is Wimpere (eye-lashes) in the Berne 
dialect?" She replied, "Well, do you mean this (eye-brows) or this 
(eye-lashes)?" She had a clear knowledge of the distinction between 
the two referents, but her knowledge of language was insufficient. 
Jaberg calls this Hnguistic uncertainty. The other case is illustrated 
by the impossibility of saying exactly where the line is to be drawn 
between cheek and chin, or between hip and thigh. Jaberg calls this 
objective uncertainty. (Cf. also Tappolet, Meringer, v. Wartburg and 
others, quoted by Jaberg; and Sandfeld, Sprogvid. 81 sqq.). 

1) K. O. Erdmann, in a well-known and often quoted passage, makes the follow- 
ing distinctions with regard to the meaning of a word: "(i) Der begriffliche Inhalt 
von grosserer oder geringerer Bestimmtheit; (2) der Nebensinn; (3) der Gefiihls- 
wert. Und ich verstehe unter dem Nebensinn alle Begleit- und Nebenvorstell- 
ungen, die ein Wort gewohnheitsmassig und unwillkiirlich in uns auslost; unter 
Gefiihlswert oder Stimmungsgehalt alle reaktiven Gefiihle und Stimmungen, die 
es erzeugt" (K. O. Erdmann 107). 

If we study Erdmann's preceding analysis, especially p. 105, we shall find 
that der begriffliche Inhalt is an attempt to define the constant element in the 
meaning of any word — an attempt that cannot succeed on these lines, as I have 
shown in 3.23. Erdmann instances the two words Leu and Lowe, which, he says, 
have identical begrifflicher Inhalt, but different Nebensinn and Gefiihlswert. The 
case is evidently parallel to that of poor young man and wretched wight, described 
above; that is to say, the two words have approximately identical referential 
range, but different semantic ranges. They refer to identical referents through 
different subjective apprehensions. 

Erdmann ascribes to the word the power of evoking Nebensinn and Gefiihlswert. 
This cannot be true of the word as a mere sound-group, but only of the compre- 
hended word, and it seems reasonable to assume that the associations are con- 
nected on the one hand with the referent (which, as shown in 4.154 sq. is an im- 
portant source of emotive colouring), on the other hand with the subjective ap- 
prehension expressed by the word, i.e., with the meaning. I do not know if it 
is really possible to make any tenable distinctions on this point. It may be added 
that Erdmann does not reckon with the speaker, for whom the emotive colour 

64 GUSTAF STERN 4. 1 7 

It is evident, as Jaberg points out, that such uncertainty may lead 
to sense-changes. Thus, the descendants of Lat. coxa 'hip', in the Ro- 
manic languages often mean 'thigh' (for instance Fr. cuisse. Meringer, 
Worter u. S. Ill 49 sqq.). Wartburg has shown that the notions 
connected with the word borgne vary in different locaHties. Jaberg 
concludes, on the basis of his own researches, that the abundance of 
synonyms in lexicological border-districts leads to Unguistic uncertainty, 
the referents themselves still being clearly distinguished. 

In my terminology, the former t5^e of uncertainty is a lack of know- 
ledge of the range of the words, the latter is lack of knowledge con- 
cerning the referent. 

Marty (Unt. 527 sqq.), in discussing what he calls die Verschwommen- 
heit der Namen, makes a similar distinction between names that are 
intentionally given a certain vagueness, and others which in them- 
selves are what he calls "unscharf". To the former group, which is equi- 

due to his adjustment to the referent is prior to the words, the selection of which 
it determines. 

The definition of Nebensinn, although limited by the terms gewohnheitsmdssig 
and unwillkurlich, does not provide an adequate delimitation against mental con- 
text. The instance of Harmodios is a case in point. Some of the notions men- 
tioned above as elements of context might fall within the meaning of the word 
according to Erdmann's definition, a consequence which I do not think he would 
accept. Erdmann has failed to see the difference between elements that are 
peripheral in the meaning of a word, and those that are peripheral in a person's 
total mental content. The two may perhaps coincide, but are probably more often 
distinct. He has also failed to see that emotive elements may be central in mean- 
ing, and thus to keep apart the two distinctions central/peripheral, and cognitive/ 

Wellander (Studien I 41) rejects Erdmann's distinction between Nebensinn 
and Gefiihlswert, which, in his opinion, are so closely connected that it is often 
"unmoglich, und, ich mochte sagen, unrichtig, jedenfalls unpraktisch" to dis- 
tinguish them. Wellander prefers to take them together as "Assoziationsgehalt", 
which would then be "jene spezifische Zugehorigkeit des Wortes zu einem bestimm- 
ten Vorstellungskreis, welche gerade diese assoziativ verbundenen Vorstellungen 
dem Bewusstsein nahe bringt". This sphere may be emotive or non-emotive. 

It would, I think, be more correct to say that the Assoziationsgehalt is "die 
verbundenen Vorstellungen" themselves, since we are dealing with psychic ele- 
ments, not with abstractions, as Zugehorigkeit. But, as I have shown, to extend 
the limits of meaning in this way leads to preposterous results. Like Erdmann, 
Wellander fails to see that emotive elements may predominate in meaning, and 
he is, as far as I know, the only writer who denies the essential importance of 
the distinction between cognitive and emotive elements. 


valent to Jaberg's linguistic uncertainty, belong expressions prefaced 
by about: about a hundred; forms like greenish, longish, or with the quali- 
fications resembling, not tmlike, related to, comparatively, and so on. The 
other type is equivalent to Jaberg's objective uncertainty, and com- 
prises adjectives like great, small, young, old, swift, white, black, grey, 
etc., as well as many words denoting ethical notions. We find this 
characteristic everywhere in names of referents permitting a greater 
or smaller quantity, or degrees of a quality, in a literal or a metaphorical 

The two kinds of vagueness correspond to two of the three factors 
of meaning, the referent and the word. It is only to be expected that a 
third kind should occur, of subjective origin, connected with the third 
factor of meaning. It was noted by Paul (Prinz. 25) "dass eine grosse 
Menge von psychischen Vorgangen sich ohne klares Bewusstsein voll- 
ziehen", a circumstance which Paul rightly considers of the greatest 

Spearman (155 sqq.) discusses at length the problems connected 
with degrees of clearness, and comes to the conclusion that the concept 
of clearness includes two items, intensity and determinateness. They 
appear to be disparate. "A startling gulf between the two reveals itself 
in the fact that degrees of intensity are applicable to realities whereas 
degrees of determinateness appertain solely to mental objects as such" 
(1. c. 159). "There appear to be numerous cases where the intensity 
is great, although the determinateness is small. Such are afforded by 
the apprehension of new notions; or even of old ones, if sufficiently dif- 
ficult. Fatigue, also, can reveal striking contrasts in this respect; over- 
night, certain items may be apprehended with great, even obsessing, 
intensity, and nevertheless obstinately remain very indeterminate; on 
the following morning, these same items may arise in consciousness no 
more intensely than before, but now with the determinateness of a 
line-engraving. Similarly, when a rather subtle argument is examined 
over and over again until understood with faciUty, the later rehearsals 
of it may quite well fall short of the earlier ones in intensity although 
surpassing them in determinateness. Or take the case where a man is 
deliberating; his apprehension of the chief relevant facts is apt to sink 
to a very low intensity as compared with various mere details, and yet 
to remain highly determinate" (1. c. 160). 

It would appear that linguistic and objective vagueness are both 

Goteb. Hogsk. Arsskr. XXXVIII: 1 

66 GUSTAF STERN 4. 1 7 

forms of indeterminateness. I am uncertain whether subjective vague- 
ness is always lack of intensity, or whether it is a form of indeter- 

Connected with these matters is the problem of conscious and 
unconscious. "Can an experience actually occur — as held by Plato, 
Plotinus, lycibniz, and perhaps the larger half of the moderns — 
without an awareness of it likewise occurring?" In favour of this 
view Spearman adduces the fact that "to introspect one's experience 
may under some conditions become exceedingly difficult". Inatten- 
tion, fatigue, want of practice, inferior native ability, and fugitiveness 
of the data, may be the causes of such difficulty. But awareness 
of experience is "not in the least obstructed by indeterminateness; 
one can be quite sharply aware of thinking unsharply". On the other 
hand, the eduction of relations and correlates^) suffers from indeter- 
minateness: "the less any fundaments are determinate, the worse will 
their relations be educed, and similarly as regards educing correlates". 
But if intensity becomes very low, "the power of experiential intro- 
spection rapidly declines, but yet that of eduction seems to remain little 
affected. In this way, it comes about that a cognitive item of low 
intensity can be very "clear" in the sense of intrinsically determinate 
and eductively effective, and nevertheless at the same time very 
"unclear" in the sense of unamenable to introspective apprehension" 
(1. c. 165—166). 

These principles seem to me to be directly applicable to the following 
experiments. During a series of tests, I read to the observers a Swed- 
ish sentance: "Med pass menar man en sanka i bergen, i vilken man 
anlagt en vag" (literally, 'with pass one means a depression in the moun- 
tains, in which one has built a road'); I then asked what meaning the 
testees attached to the word man which occurs twice in the passage, 
with different meanings; the first time 'people in general', the second 
time 'some people who have built the road'. It was found that the 
word had not been noticed. Its function, if we explain it according to 
Spearman's theory, is to be a fundament of certain syntactical relations. 
Another testee had not consciously noticed the word menar 'means'; he 
had merely apprehended the chief notions of the sentence: pass, de- 
pression, etc., but he nevertheless related them to each other quite 

') On these terms, see note to 4.265. 


It seems as if a large portion of what we actually hear or read is not 
consciously apprehended; it remains at a very low degree of intensity, 
but nevertheless it is sufficiently determinate to guide our mental pro- 
cess; we interpret the sentence correctly with regard to all its consti- 
tuent parts. And in spite of the low intensity, we are very sensitive to 
discrepancies. If I had made the very slight modification of intro- 
ducing the Swedish pronoun han 'he' instead of man 'one', the testees 
would no doubt have noticed that there was something wrong. 

I shall have occasion later on to speak of the supplementation of 
relations on the base's of apprehended fundaments: we do not in listen- 
ing or reading notice carefully every ending and form-word, but educe 
the relations which they denote on the basis of the fundamental words 
of a speech. But if the endings are wrong we quickly notice the fact. 
In Spearman's terminology, these items are of a very low intensity, and 
therefore difficult to introspect, but they are sufficiently determinate 
to make them capable of functioning as members of a relation. 

Stout makes similar statements: "It is certainly possible to think 
of a whole in its unity and distinctness without discerning all or even 
any of its component details". Further, there are instances "in which 
only a certain part or certain parts of the whole are distinctly apprehend- 
ed while the rest appears as a distinctionless unity — as a somewhat 
which may be separated into its component items, but which is not 
so separated at the moment. It is no exaggeration to say that this mode 
of thinking pervades our whole mental life" (Stout I 78 sqq., 92. Cf. 
Burkamp 248, G. E. Miiller III 513, Moore, Mono. 255, 276, Ward 307). 

A large portion of the psychic content that constitutes the meanings 
of words thus comes to lie "in the twilight of language" (Wilh. Ander- 
sen), where changes may happen without being noticed. If all men- 
tal experiences were equally intense and determinate, any modification 
of meanings would be more likely to be noticed and corrected. As it 
is, discrepancies with regard to some elements may pass unnoticed if 
only they are able to fulfil the functions that devolve on them a 
fundaments for relations; that is to say, they may have a very low 
intensity, if they are only sufficiently determinate. 

The correlation between the distinctions here made, and those spok- 
en of in the preceding paragraph, are obscure. It would seem, however, 
that peripheral elements are weak both in intensity and deter minateness. 
Whether these two qualities fail simultaneously, or if sinking intensity 

68 GUSTAF STERN 4. 1 7 

is the first step from central to peripheral, is a problem that would 
require further investigation. 

It seems probable that it is especially the lack of determinateness 
that paves the way for sense-changes "in the twilight of language", 
lyinguistic and objective vagueness involve a certain elasticity of the 
objective and semantic range, owing to lack of definite limits. If a 
border is vague, it may easily be over-stepped. 

4.2. Logical Elements and Categories of Meaning. 

4.21. Introductory. The problems to be dealt with in the present 
section have attracted much more attention and been much more 
discussed among philologists than those pertaining to the psychic con- 
stituents of meaning. Nevertheless, many aspects of the matter are 
still very inadequately known, and require much additional work be- 
fore we can say that the problems are stated with precision. Every 
one of the distinctions described below is in need of further elucidation, 
and my exposition is only a preliminary survey of the field. Much 
confusion has been caused by Paul's unfortunate definitions of usual 
and occasional meaning, which have been repeated in later handbooks 
in spite of the criticism long ago directed at them by competent schol- 
ars. It will be apparent in the following chapters that many sense- 
changes cannot be satisfactorily explained if we do not make use of 
the fundamental distinctions established by Husserl and other writers. 

4.22. Actual and Lexical Meaning. Isolated Words. Actual meaning 
is the meaning of a word in actual speech. I^exical meaning is the 
mental content attaching to an isolated word (or phrase). When I 
say "isolated", I do not mean merely syntactically isolated, as a word 
functioning as a sentence {Fire!), but a word that has really no con- 
text, external or psychic; for instance the words in a dictionary. A 
sentence used in a grammar to illustrate a rule {e. g., the candle was 
burning brightly) is also isolated in this sense (Cf. Pos 59, 61). 

In ordinary speech isolated words and sentences do not occur. The}' 
are a peculiarity of the study of language. "We never — except for 
the sake of this very inquiry — attempt to fix our minds . . . upon 
some isolated concept; in actual thinking ideas are not in conscious- 
ness alone and disjointedly, but as part of a context. When the idea 
'man' is present, it is present in some proposition or question, for 


example, Man is the paragon of animals, In man there is nothing great 
but mind, Is man immortal? and so on" (Ward 299).^) 

The present study is concerned only with actual meanings. The 
meanings of isolated words are discussed here in order to demonstrate 
the distinction between them and actual meanings, a distinction that, 
has not always been sufficiently recognized, nor correctly analysed. 

The first essential difference between actual and lexical meaning is 
that the latter lacks a definite objective reference, which, as shown in 
the preceding chapter, is a constitutive characteristic of actual meaning. 
This comes out clearly if we compare the sentence quoted, the candle 
was burning brightly, as used in a definite context, in actual speech, 
with the same sentence occurring in isolation. In the former case, 
we know more or less about the context, the situation, and we are 
able to put the import of the sentence into relation with our previous 
knowledge, and to apprehend the actual referent which it denotes. If 
the sentence is isolated, such a mental operation is impossible. We 
know of course what a candle is, and we know what it is to be burning 
brightly, but we do not know what candle it is that is burning brightly, 
or where, or when, or why: we are not able to assign to the candle and 
its burning a definite "place" in any perceived or imagined universe. 
We are able to imagine possible referents for the phrase, but we cannot 
fix upon any one of them as being that intended by the speaker — 
because there is no speaker, in the sense in which I have taken 
that word. This is the second essential difference between lexical and 
actual meaning. The former is not an expression of a language-user's 
subjective apprehension of a referent, because an isolated word is not 
used in actual speech and has no definite referent. An isolated sentence 
is a proposition, but not a judgement. It lacks the purposive function 
characteristic of speech, because there is no speaker whose purpose 
it could express. 

This applies even to so-called unica. The moon! spoken without any 
context, external or mental, to indicate its setting, will make the hearer 
ask: "The moon? Well, what about it?" A word has actual meaning 
only in a sentence or by suggesting a sentence (Bosanquet I 37 — 38, 
Pos 61, 67 — 68, Gardiner, passim; and 6.31 below). 

Paul and other writers have confused the distinction between lexical 

^) Humboldt § 32: Die Sprache liegt nur in der verbundenen Rede. Gram- 
matik und Wort^rbucU sind kaum ihrem todten Gerippe vergleichbar. 


and actual meaning with the distinction between general and particular 
meaning, which is discussed in the next paragraph, and especially 
have they confused the two notions lexical and general meaning. They 
assume that the lexical meaning of a word is equivalent to its most 
general actual meaning; that, for instance, an isolated animals would 
have the same meaning as the word animals in the sentence all ani- 
mals were created by God. This is a mistake. In the latter case the 
word has a definitely "placed" referent; in the former a reader or listen- 
er knows of no such definite reference: "we are set making proposi- 
tions at randon" (Bosanquet). Lexical meaning has also been confused 
with the range of the word (cf. 4.3). 

As will be shown in 6.22, the mental content associated with an 
isolated word, leaving aside the feeling of familiarity or strangeness, 
is composed mainly of the following items: (i) an awareness of the 
sphere or direction in which the referent is to be sought, or, if the 
word has many dissimilar meanings, possibly of more than one sphere; 
(2) an awareness of one or more of the applications of the word; (3) 
images representing one or more of the possible referents of the word. 
More briefly, we might express this in the following way: an aware- 
ness, more or less clear, of the range of the word, and of a smaller 
or greater number of items falling withing the range; images may also 

On the whole, if we define meaning as I have done, it is not correct 
to speak of the meaning of an isolated word. Two of the three deter- 
mining factors of meaning are left vague: we do not apprehend defi- 
nitely the objective reference, and we do not know under what aspect 
the referent is to be apprehended — the subjective apprehension is 
vague. It is only the third determining factor, the traditional range, 
that is, or at least may be, definitely known. 

Lexical "meaning" should thus be defined more precisely as an 
awareness of the range of a word. Such an awareness is probably in 
most cases not much more than a rather vague awareness of direction, 
in which one item or another, arising at random, may stand out in 
greater clearness, either in the form of thoughts or of imagery. 

4.23. General and Particular Meaning. Compare the following 
phrases, (i) the dog is a domestic animal, and (2) that dog is mad. In 
the first instance dog has a collective, general import, equivalent to 
'all dogs'. In the second instance, dog refers to one particular ani- 


mal.^) In this way all appellative nouns can be used either to represent 
the species or to denote an individual (in the plural a number of indivi- 
duals) belonging to the species; in other words, it can be used in a 
general or in a particular sense. 

The same is true, with slightly varying characteristics, of other 
nouns, of adjectives and of verbs. We may speak of running as a use- 
ful exercise, in the general sense of the verb, or we may speak of my 
dog that ran away, in a particular sense. Husserl (II. 1.87) gives 
another instance: es gibt Kitchen (hie et nunc), and es gibt regelmdssige 

When a word is used in its particular sense, the referent is one 
or more items within the referential range. The referent of the 
word used in its general sense is not easy to define. The dog cannot 
denote the general concept or category of dogs, for it is not the concept 
or the category that is a domestic animal. To interpret the sentence, 
'the category of dogs belongs to the category of domestic animals' 
seems to miss the point of the statement, which is to ascribe to real 
dogs a certain characteristic. The word in the singular appears to 
represent in a peculiar manner all and any dogs; or is it the one dog 
denoted that represents all his fellows? If we turn to abstract nouns, 
as liberty involves responsibility , the problem has yet another aspect. 
I have to leave this very knotty point to the logicians. — At any rate, 
the general concept cannot be the meaning of the word (cf. 3.15).^) 

1) A generalizing or a particularizing function is sometimes ascribed to the 
definite article in English. It is of course the noun, not the article, that thus 
varies in meaning. 

^) Writers who regard the general concept as the meaning of the word in its 
general sense have to discuss its formation and characteristics. I take the ex- 
istence of general concepts for granted; to analyse their nature and origin is a prob- 
lem for epistemology. Some philologists at least appear to explain general con- 
cepts only through abstraction, a view for which I find no support among logi- 
cians and psychologists. "II est impossible de reduire la generalisation a I'ab- 
straction, qui ne saurait conferer la generalite au caractere qu'elle isole, ou a la 
substitution qui, loin de fonder la generalite, la presuppose" (Delacroix 90). "To 
describe the process by which "insight" is obtained as a mere matter of abstrac- 
tion, the result of association . . . deserves the stigma of 'soulless blunder' which 
Hegel applied to it" (Ward 304). Spearman 266 sqq. gives a review of some ex- 
periments on the problem. See also Gomperz II. 1. 169 sqq., 228 sqq.; Cassirer. 
Spr.u.M. 20 sqq., Phil. I 244 sqq.. Ill 134 sqq., 335 sqq., 365 sqq.; Moore, Mono, 
238, 242, 262 sqq.; Ogden- Richards 154 sqq., and especially WillwoU 4 sqq. 


4.24. Specialized and Referential Meaning. In the sentence, / 
can just make out that one of the two is a man, and the other a woman, 
the speaker is thinking, in connection with the word man, almost 
exclusively of exterior characteristics, dress, etc., that show a human 
being to be a man {vir); in He was a man, take him for all in all, man 
refers primarily to certain moral and mental qualities that are con- 
sidered as typical of a man of honour; in Man is immortal, the word 
refers to 'human being' in general {homo); and in He had an army of 
ten thousand men, men refers to the soldiers as entities, without signi- 
fying any specific characteristics (Cf . the instance of the hrick adduced 
in 4.16 above; Ivipps, Wortbedeutung 61, Bain I 203 sqq., Greenough 
& K. 248 sqq.). 

Another instance is What a child you are! where child means 'childish'; 
that is to say, the meaning of the noun is specialized to denote certain 
characteristic qualities of the referent. In this way a noun may receive 
adjectival function. (Cf. Sweet, Grammar § 173; Jespersen, Gr. a. Str. 
134; Efvergren 47). 

In general terms, if a referent has the characteristics a b c d, a speak- 
er's attention may be directed to one of them, or to a combination 
of them, and need not embrace the whole complex (cf. Scripture 63 
— 64). I call this peculiarity the specialization of meaning. For one type, 
the last of the four mentioned with regard to man, in which the speaker 
refers to the referents as entities without attending to their charac- 
teristics, I use the term referential meaning. Referential meaning is 
thus one kind of specialized meaning (cf. Conrad, Arch. 19, 453 sqq.) 

From the psychological point of view we have to notice that when 
complex psychic structures are associated, bonds are formed also be- 
tween elements, not only between the totals. It is thus in accordance 
with known facts if we assume — in explanation of the phenomenon of 
specialization — that the word is associated with the apprehension of 
each characteristic of the referent, not only with the apprehension of 
the referent as a whole. Note that each characteristic is of course appre- 

It is evident that the word plays a very important part in the formation of 
general concepts (Cassirer, Phil. Ill 135). It is the most substantial item of the 
complex, round which the other elements group themselves. Jacobsson (Begrep- 
pet 205) defines Begriff 'notion' as a precisely determined unit of meaning. — I 
have already spoken in the preceding paragraph of the frequent confusion of 
lexical and general meaning. 


bended as inherent in the referent in question, not as an isolated item. 
The reference to the whole entity should be assumed to be always 

There is thus no difficulty in explaining how the apprehension of, 
for instance, any one characteristic of a man can always evoke the 
word man. Conversely, in the hearer, the word is able to evoke that 
aspect of the referent which is relevant to the context, leaving the 
other aspects outside the consciousness.^) 

In a word like brick we do not speak of different senses or shades of 
meaning. In other cases, one or more aspects of the referent gain in 
importance and independence, and new senses of the word arise {man 
= 'vir' or 'homo'). Sometimes one language will, as in the instance 
given, retain the same word for different aspects of the referent, while 
another language marks the distinction by the use of different words. 
The importance of specialization as a starting-point for sense-changes 
is obvious and will be constantly mentioned in the second part of 
this book. Cf. Stocklein 9. 

Specialization is not confined to cognitive elements. In / have oceans 
of time, the cognitive import of oceans is altogether lost, and there 
remains only the feeling of something very big (cf. 14.531). 

The number of definitions of specialized meanings in a dictionary 
are an indication of the varying aspects of the referent denoted by the 
word in question. The referent, for instance, of an adjective denoting 
a simple quality, is almost homogeneous, and there will be very few 
shades of meaning to define. 

If we compare specialized meaning and particular meaning, we find 
that primarily the former involves a restriction to a part of the seman- 
tic range of the word, the latter a restriction to a part of the referential 
range (3.25). Cf. the instances given above. We find, further, that a 
general meaning is also specialized, as in Man is immortal. On the 
whole, it is the normal thing for a meaning to be specialized, whether 

1) Cf. Thorndike, Psych. II 34: "When such a part {i.e., part of a total situation) 
happens alone (It really never happens alone, being always a part of some total 
state of affairs. The alone means simply that it is a very distinct and predominant 
element of the total situation) or in a new context, it does, as was stated under 
the laws of partial activity and response by analogy, what it can. It tends to 
provoke the total response that was bound to it; it tends especially to provoke 
the minor features of that total response which was especially bound to it". Cf. 
below 1 1.5 1. 


it is general or particular. It must be very rare for the total semantic 
range of a word to be simultaneously present in mind. 

We may, further, compare specialization with the distinction be- 
tween central and peripheral elements of meaning (4.16). If we special- 
ize on any one characteristic of the referent, that characteristic is 
central in meaning, and the two notions speciahzed and central are 
thus correlative, the former pertaining to the logical, the latter to the 
psychological point of view. 

4.25. Tied and Contingent Meaning. Husserl (II. i. 79 sqq.) dis- 
cusses what he calls, on the one hand, wesentlich ohjektive Ausdriicke, 
and on the other hand, wesentlich subjektive und okkasionelle Ausdriicke. 
The latter show a peculiar shifting of the actual meaning, which Hus- 
serl illustrates by the phrase / wish you hick. It is evident that the re- 
ferent of this phrase changes from case to case; innumerable different 
persons and different kinds of success may be denoted. But this poly- 
semy is quite different from the polysemy, for instance, of the German 
word Hund, signifying either 'dog', or 'kind of cart used in mines'. It 
is especially the latter type that is meant when writers speak of equi- 
vocation; but it would be possible to avoid the equivocation, while in 
the case of / wish you luck it is unavoidable. I shall use the terms tied 
and contingent meaning. 

A meaning is a tied meaning, according to Husserl, when its reference 
is determined by word and verbal context, without regard to the person 
speaking or to other circumstances. A tied meaning may also vary: 
there is then more than one possibility, each determined in the man- 
ner stated. 

A contingent meaning is not absolutely determined by word and 
verbal context; it shifts with the momentary external context, including 
the speaker. The hearer must take these into consideration in order to 
interpret the words correctly. 

It seems to me that the distinction lies essentially in the circumstance 
that a tied meaning refers to a category of objects, or to one or more of 
the individual objects belonging to such a category. The word may be 
able to refer to more than one category, as in the case of Hund. A 
contingent meaning, on the other hand, is able to refer to a variety of 
referents that do not together form empirical categories. 

The meaning of all theoretical statements, statements concerning 
principles and doctrines, demonstrations and theories of "abstract" 


science, are tied. The actual circumstances have not the sUghtest in- 
fluence on the meaning of, for instance, a mathematical thesis. We 
read and understand it without thinking of the speaker or writer. Ex- 
pressions in daily life, on the other hand, constantly show contingent 
meanings. This is the case with any sentence containing a personal 
pronoun. The word / denotes different persons on different occasions, 
and the correct reference is to be gathered only from the external con- 
text, the circumstances of the utterance. 

What applies to personal pronouns, applies also, with modifications, 
to other pronouns, and to words like here, there, above, below, now, yes- 
terday, to-morrow, afterwards, and so on. The contingent character of 
such meanings is transferred to all sentences in which the words are 
used, and these include all utterances in which the speaker expresses 
anything concerning himself or concerning things thought in relation 
to himself; that is to say, all expressions for perceptions, beliefs, doubts, 
wishes, hopes, fears, commands, and so on. Here belong also all nouns 
with the definite article, referring to individuals that are denoted only 
by the name of the genus or class. When we speak of the king, we mean 
our own king, when we ask for the lamp, we mean our own lamp (Hus- 
serl 1. c). Jespersen (Ivanguage 123) employs the term shifters for 
words with contingent meaning. We shall have occasion to recur to 
this matter in discussing the unique use (14.85). 

4.26. Basic and Relational Meaning. 
4.261. Introductory Remarks. Meillet has called attention to the 
fact that in the earlier Indo-European languages there is no form that 
represents the noun as such without regard to the relations into which 
it necessarily enters when used in speech. In Latin, for instance, it 
was necessary to say lupus, or lupi, or lupos, and so forth, and there is 
no form without the endings which always assign to the word a de- 
finite syntactical function (Meillet, J. de. Ps. 20, 246, also quoted De- 
lacroix 8 note; Amman, Rede I 34). Wundt had previously (Die Sprache 
I 594) made a distinction between Grundelemente und Beziehungsele- 
mente des Wortes. "Grundelemente nennen wir hier wieder diejenigen 
Lautbestandteile, die fiir den innerhalb einer bestimmten Wortgruppe 
konstant bleibenden Begriff characteristisch sind, wahrend die Bezieh- 
ungselemente solche Bestandteile umfassen, durch die jener Begriff 

76 GUSTAF STERN 4.261 

irgendwie modifiziert und dadurch zugleich zu andern in die Rede 
eingehenden Worten in Beziehung gebracht wird". However, Wundt 
did not apply the distinction to meanings, but only to the form of words, 
and he has not worked it out in detail.^) 

In many languages syntactical relations, for instance the genitive 
relation denoted by the -i of lupi, are denoted by particular words; in 
other cases they are left undenoted or are indicated by word-order. 
Thus there is nothing in the form of the two proper names in Jack beats 
Jill, and Jill beats Jack, to show the different syntactical functions (cf . 
Wundt II. I sqq.). 

I shall make use of the terms basic and relational meaning to distin- 
guish the elements of meaning corresponding to the stem of a word from 
those corresponding to certain (not all!) endings. The relational mean- 
ings are relations in speech, and their fundaments are the basic mean- 
ings. The referents of these two kinds of meanings are, respectively, 
relations holding between referents, and the fundaments of these rela- 
tions, i. e., the basic referents. 

As stated in ch. i, the relational meanings are not exhaustively treat- 
ed in this investigation, but it is necessary nevertheless to define the 
boundary between basic and relational meanings, and especially to 
discuss the combination of basic and relational meanings, as well as 
some peculiarities of the latter. 2) 

4.262. The Psychic Reality of Relations. Relations as Referents.. A 
few words should perhaps be said on the importance of relations in 
speech, and on their psychic nature. For the first point I refer to the 
following chapter, where abundant evidence will be given regarding the 
importance of relations in the production and comprehension of speech, 
as shown by Head and other writers. 

^) The words carrying an independent meaning "sind . . . jene, die Aristoteles 
die kategorematischen Redeteile nannte: Haupt-, Fiir-, Eigenschafts-, Zeit-, Um- 
standsworter u. dgl. Die Verbindungsworter dagegen — und ihnen treten gleich- 
berechtigt die grammatischen Formen sowie die Wortstellving und die Unter- 
scheidungszeichen zur Seite — bezeichnen, da sie eine selbstandige Bedeutung 
nicht haben (und darum nennt sie ja Aristoteles synkategorematische Redeteile), 
nicht "Telle" des durch Zusammenfassung zu gewinnenden Bedeutungsganzen, 
vielmehr schon die zu wahlende Art dieser Zusammenfassung selbst." (Gomperz, 
Sinn 45). 

^) Cf. on this subject also Otto 60, B. Erdmann II 371, Bosanquet I 19, 
Delacroix 200, I,indroth 133, Svanberg 81 sqq., and especially Salomaa. 


With regard to the second question, opinions among psychologists 
seem to differ. The meaning of the Beziehungselemente of a word, ac- 
cording to Wundt (I 595) "besteht. . . nicht in einem selbstandig zu 
denkenden Begriff, sondern in einer begriffhchen Beziehung, die zu 
ihrer realen Vergegenwartigung im Bewusstseim immer der Verbindung 
mit Grundelementen bedarf". He admits, however, that "da diese Be- 
ziehungselemente mit ahnlich sinnmodifizierender Wirkung in den Ab- 
wandlungsformen anderer Worter ebenfalls vorkommen, so besitzen 
auch sie eine relativ konstante Bedeutung". An extreme case is repre- 
sented by the abstract particles, which denote a relation and nothing 
else. They express only "eine unbestimmte Beziehung, die isoliert 
nicht vorgestellt werden kann. Hier wird das Wort im allgemeinen bloss 
als Wort vorgestellt, als gelaufiger L^autkomplex, der sich aber vermoge 
der gewohnten begriff lichen Anwendung mit einem Gefiihl verbindet, 
das wahrscheinlich von andern, haufig mit ihm verbundenen Wortvor- 
stellungen ausgeht, die sich assoziativ zum Bewusstsein drangen. . . 
Das reine Beziehungswort erweckt zunachst nur eine lyautvorstellung, 
an die irgendein Gefiihlseindruck gekniipft ist, der gelegentlich durch 
wechselnde aussere Wortassoziationen abgelost werden kann" (1. c. 


Bourdon (200) expresses similar opinions. If words like done, si, 
mais, have any meaning when taken in isolation, the reason is that they 
are associated to mental attitudes, and not that they denote relations; 
si is associated to a more or less pronounced attitude of hesitation or 
doubt, mais to an attitude of reservation, and so on. Bourdon quotes 
experimental results by Michotte and Ransy (Contrib. a 1 etude de la 
memoire logique 13 and 29) which, however, do not appear quite con- 

In criticism of these views it should first be pointed out that Wundt 
and Bourdon do not employ the theory of imageless thought, and there- 
fore scarcely have any other resort than feelings, when they cannot 
analyse a mental content as constituted by imagery. Secondly, the 
isolated meaning of the particles proves nothing with regard to the 
actual meanings. 

If we turn to actual speech, we shall find that endings and words (as 
well as word-order), as expressing relations, are used regularly, in an 
identical manner, in the most varying contexts. In analogical forma- 
tion of flexional and derivational forms, the relations are handled — 

78 GUSTAF STERN 4.262 

from a very early age — exactly as their fundaments are, and their inde- 
pendence does not seem to be less (9.22). We cannot reasonably ex- 
plain this in any other way than by attributing to these endings and 
words a constant meaning, which they express whatever the context. 
This is an empirical fact, and it agrees with Spearman's statement that 
"any apprehended relation is in itself an item in the cognitive field over 
and above its apprehended fundament" (Spearman 158). From this 
cognitive import of the particle we should distinguish the emotive ele- 
ments which may go along with it, as for instance in mats and si, 
adduced by Bourdon. But I do not see how the meaning of and 
could be explained in this way. 

Whether it is possible to conceive a relation in isolation, as Spearman 
contends, although of course in actual fact no relation (as referent) 
occurs except as holding between fundaments, — is not relevant for us 
at present. (See on this point Husserl II. i. 306 and 313 sqq., 
apparently concerning isolated words; Bichowsky; Schwarz 160, Biihler 
Arch. 12, 9 sqq.). 

I propose, then, to regard relations as the referents of certain 
endings and particles, functioning in the same way as other referents 
with their various names. The apprehension of the relation is the 
basic meaning of the particle or ending. 

Since every word in a sentence stands in a syntactical relation to one 
or more other words, and these relations constitute the relational mean- 
ings of the words, it follows logically that a particle has a relational 
meaning, which is the actual relation that, in the case of and, for 
instance, holds between the notion of togetherness and the meanings 
that are together. Spearman calls this the attributive relation: "It 
includes, for instance, the relation of a character to its fundament, as 
of redness to the thing that is red. Another instance is the relation 
borne by any relation itself to either of the things related, as that of 
fatherhood to father" (Spearman 69. Cf. also Biihler, Arch. 12, 10, 
quoting Meinong). This relational meaning of a particle is always left 
unexpressed. 1) 

4.263. Syntactical Relational Meaning. It will be useful to begin 

^) The reality and independence of the meaning of a stem, a basic meaning, 
is of course empirically proved by similar arguments. Cf. Pos 71. See also Will- 
woU 135 sqq. on the importance of relations in our mental activities, and Salomaa 
165 sqq. 


the discussion of the various kinds of relational meanings with the 
simple and obvious case of stem and flexional ending, as for instance 
the genitive lupi. To linguistic instinct, the stem lup- represents the 
notion of the referent as such; the ending -i the genitive notion, which 
corresponds to a relation between the wolf and some other referent. 
The verbal form amavi represents another temporal relation than amo, 
and, in comparison with the passive amor and amahis sum, other rela- 
tions with regard to subject and object of the action. The type of re- 
lational meaning denoted by these endings may be called syntactical 
relational meaning, since it corresponds, from the semantic point of 
view, to the syntactical function of the word. 

Flexional endings and syntactical relational meanings are by no 
means always correlative. Many flexional endings denote, not relations, 
but modifications of the basic referent. This is the case with dual and 
plural endings, and apparently with the comparative and superlative 
endings of adjectives and adverbs, although these also have a relational 
character. Subjunctive endings (or other modifications equivalent to 
such endings; see below) indicate a subjective attitude of the speaker 
towards the referent, which I do not include under the heading of rela- 
tions; similarly the imperative denotes an attitude towards the listener 
(see 4.15 above). Tense endings denote relations in time, and belong 
here; similarly passive endings, which denote a re-arrangement of the 
relations subject-action-object, as compared with the active. (Cf. Pick, 
Sprachstor. 119, quoting Riess, Was ist Syntax?). 

On the other hand, syntactical relations are not always denoted by 
endings. Sometimes they are denoted by other modifications of 
form, as for instance in ring, rang, rung; sometimes by form-words, 
and sometimes they are not denoted by any verbal form, but either by 
word-order or not at all. In this case the speaker leaves it to the 
listener to educe relations from the given fundaments, e. g.. Jack beats 
Jill, as compared with Jill beats Jack. A leather shoe is a shoe made of 
leather, but a leather varnish is a varnish intended for use on leather. 

Oertel points out that endings often have many meanings. The end- 
ing of amarem expresses (i) the attitude of the speaker, (2) the person 
of the speaker, (3) the time relation; the -i in domini combines (i) gender, 
(2) number, (3) the "case" relation, (4) an implication of the nominal 
character of the word. This is perhaps not quite correct. It seems to 
me that the meaning of -i should be regarded as a contingent meaning. 

80 GUSTAF STERN 4.263 

and that the relational meanings attributed to the word are partly 
educed from awareness of the fundaments and their connections, so 
that the comprehension is not founded only on interpretation of the -i. 
The nominal character of the word is of course Unked up with the word 
as a whole, not with the ending, since -i may also be a verbal ending. 

It is very common for syntactical relational meanings not to be de- 
noted by any word or part of a word. Language apparently cares little 
for relations and often employs expressions that may be logically inex- 
act but are more effective, brief and to the point: how charming she is 
with her dark curls; a logically complete expression would spoil the ef- 
fect. (Further instances see Lerch, GRM 1913, Biihler, Ber. 120 sqq.) 

4.264. Derivational (Relational) Meaning. Another type is repre- 
sented by the meanings of prefixes and suffixes as modifying the mean- 
ing of a stem, the basic meaning. In the group like — liken — like- 
ness, there is a basic similarity of meaning which is modified in various 
ways in the derivations (Cf. on derivational groups, 9.25). The 
suffixes -en and -ness are clearly apprehended as having a definite 
meaning of their own, since they are freely used to form new combina- 
tions with new stems, and always modify the meaning of these stems 
in the same way. Sometimes the variation of function finds no 
expression in the form of the word, as in round, which may be a noun, 
an adjective, a verb, an adverb, or a particle; a multiplicity that has a 
uniting link in the cognate basic meanings. 

A typical difference between derivational meanings and the syntac- 
tical relational meanings discussed in the previous paragraph, is the 
circumstance that the latter correspond to the syntactical functions 
of the word in the sentence. That is not the case with derivational 
meanings. A "passive" noun, as examinee, may very well function as 
subject (agent) in a sentence, without losing its passive character. 
The relations denoted by the endings are relations to some item denoted 
by the stem. (Cf. Stern, Studia Neo-phil. II 102—103). The feeling 
for this relation must be alive to the speaker at the moment of coining 
a word, for instance trustee. But if the word comes into current use, 
adequation (see ch. 14) soon sets in. The referent will always have 
other characteristics than that of being trusted: a trustee has positive 
functions, as manager of an estate, etc., and the meaning of the word 
may be variously speciaUzed, so that the relational aspect in certain 
contexts falls into the background, among the peripheral elements, or 


is even altogether lost. That is the case in committee, as shown by the 
shift of stress. The adequation is then complete. 

After adequation, the relation between the meaning of the derived 
word and the meaning of its stem is felt only in the case of specializa- 
tion in this direction. With that exception, the meaning of the word 
then belongs to a group discussed by Moore (Mono. 191); he reckons 
among the characteristics which some meanings have, though all do 
not have them, "a consciousness of relation as an integral part of it- 
self. Thus, I cannot think of creature, cause, father, etc., without re- 
ference to something else beside the creature, the cause, the father, etc. 
Any analysis of these concepts leads to their correlatives, creator, 
effect, son, etc., though I can, for instance, confine my attention to 
the concept of 'being' without, in the same sense, being obliged to 
refer to something to which 'being' is related". 

Moore is no doubt right in saying that "any analysis of these con- 
cepts leads to their correlatives", but this does not imply that we 
cannot "think of father, etc., without reference to something else be- 
side father, etc.". Moore has not taken into consideration the varying 
specialization in actual speech, when other characteristics of the 
referent may occupy the focus of attention, and the relations men- 
tioned be lost sight of. 

It seems to me that words Hke those quoted by Moore, as well as others 
{point, head, foot, big, new, etc.) in so far as the relation involved in 
their meanings is attended to, are not dissimilar to the words with con- 
tingent meanings (4.25). We do not know precisely the meaning of head 
or cause if we do not know of what it is the head or the cause. I sug- 
gest that these meanings are determined by relations to other referents, 
while in the case of contingent meanings the determination rests on the 
relation to speaker and hearer, their placing in time and space, and 
so forth. Whether the distinction is essential, or if the two classes can 
be merged, is a point that requires further investigation. 

When we have a genitive like trustee's, the meaning of which, the- 
oretically, is composed of stem meaning plus derivational meaning plus 
syntactical relational meaning, it will generally be the distinction be- 
tween the first two, taken together as basic meaning, and the third, as 
relational, that is sensible to linguistic feehng and important for com- 

4.265. "Indicated Purport". The meaning of words Hke trustee, or, 

Goteh. Hogsk. Arsskr. XXXVIII: i 

82 GUSTAF STERN 4.265 

in general, of derivations in which the derivational suffix may still be 
felt as a distinct item, are instances of what Spearman terms indicated 
purport. Purport is Spearman's term for the meaning of any com- 
bination of words, whether compound, phrase or sentence. "In this 
kind (sc. of contribution to the purport of a passage; Spearman has 
not appUed the principle to derivations), the individual meanings are 
made to serve, not on their own account at all, but only by way of 
indicating further cognitive items. For an example, we may take the 
historically interesting case of "the-root-of-two". Here, the purport 
itself, as was rightly noticed by Bolzano, is far from being exhaustively 
constituted by the meaning of the four given words. Indeed, these 
may not enter into the purport at all. To arrive at this latter is an 
achievement that requires our third principle.^) Two supplies the in- 
itial fundament; the-root-oj evidently includes a relation of the two; 
and to obtain the character of this root is nothing else than to educe 
the correlative fundament. "Indication", then, has its basis in cor- 
relative eduction . . . These two individual meanings together give birth 
to their correlate, and in the very act of so doing, they renounce being 
any portion of the purport themselves" (Spearman 120 — 121). 

This point was noticed by Biihler, who makes a distinction between 
direktes und indirektes Meinen (Arch. 9, 357 sqq.) The former is the 
awareness that I mean "this: it has such and such characteristics . . .". 
The latter is the awareness, "what I mean is that which fills such and 
such conditions". "Wir konnen auch sagen: Beim indirekten Meinen 
werde der Gegenstand durch den Akt des Meinens selbst erst gebildet, 
wahrend er beim direkten Meinen schon fertig sei und das Meinen nur 
eine Beziehung auf ihn enthalte" (1. c. 359; cf. 6.36, and Fischer, 
Arch. 43, 44). 

Spearman's analysis evidently fits the case of words like trustee: the 
stem denotes the action which serves as the initial fundament; the 

^) Spearman formulates three "noegenetic" principles, each stating one way in 
which fresh mental content is created on the basis of existing content. The sec- 
ond and third of these are of interest to us. The second principle runs as follows: 
the mentally presenting of any two or more characters (simple or complex) tends to 
evoke immediately a knowing of the relation between them. This is termed eduction 
of relations (Spearman 62 sqq.). The third principle is the principle of eduction 
of correlates, and runs: the presenting of any character together with any relation 
tends to evoke immediately a knowing of the correlative character (1. c. 91). I shall 
have occasion to quote these principles more than once in the sequel. 


ending -ee is used to denote a relation, that of being the object of ac- 
tion; the second fundament is the referent of the whole word, the person 
who stands in this relation to the action of trusting, and who is thus 
indirectly indicated. The -ee, from the Fr. participial ending -ee, Lat. 
-ata, originally a flexional ending, has changed in English into a deri- 
vational ending. Compare a word like pen-hold-er, in which both funda- 
ments and the relation are expressly denoted; and leather shoe and 
leather varnish, where the relation has to be educed (cf. Stern, Studia 
Neophil. V). 

Whether all derivations are instances of indicated purport is a prob- 
lem that requires a thorough analysis, like the whole question of rela- 
tional meanings. It should give interesting results, not least for syn- 
tactical studies. 

4.27. Word-meaning and Phrase-meaning. I make use of the term 
phrase-meaning to denote the total meaning of any combination of 
words — whether syntactically a sentence or not — as distinguished 
from the meaning of a single word. 

The functions of the phrase as symbol, expression, and communica- 
tion, sometimes also as purposive speech (cf. 2.13) are not wholly ex- 
pHcable through the functions of the single words: the phrase meaning 
generally contains supra-summative elements. I refer to 6.38, where 
the genetic point of vieW can be applied (see also 5.3, 5.5, and 6.34). 

The problem of word and sentence has been the subject of much 

The sentence is perhaps the unit of living speech, but nobody has 
yet succeeded in formulating an acceptable definition. That problem 
is outside the scope of my investigation, but something must be 
said on the interrelations of word-meanings and phrase-meaning, 
from a descriptive point of view, and in the subsequent chapters 
from a genetic point of view.^) 

It would be a mistake to suppose that the complexity of an expres- 
sion and consequently of meaning (except where adequation has inter- 

^) I refer to the following authors: Brugmann, Satzgestaltung 16, Brugmann- 
Delbriick, Vgl.Gr.3, 75, Meyer-Liibke, Gr.d.rom.Spr.3, 307, Biihler, Satz, Dittrich, 
Phil. Stud. 19, 93, Maurenbrecher 239 sqq. , Meillet, Remarques 609 sqq., Morgen- 
roth GRM 4, 5 sqq., Pos 60 sqq., Sapir 86 sqq., H. Schmitt 219, Jespersen, Ph. 
of Gr.305, W. Stern, Kindersprache 170 sqq., Gardiner Br. J. of Ps., Pick, Sprach- 
stor. 130 sqq.; J ahresbericht d. german. Philologie 1916, 3, 42 — 48; and 5.31 


vened) must necessarily imply a corresponding complexity of the re- 
ferent. As Husserl points out, simple referents can be apprehended 
through complex meanings. The instance simple referent (einfacher 
Gegenstand) is sufficient proof. Conversely, simple meanings may cor- 
respond to complex referents; Husserl considers that something and one 
{Etwas and Eins) are conclusive evidence, since these words, of course 
in the most vague and general manner, may refer to any complex re- 
ferent. Proper names are also a case in point. 

Husserl further states that even if both meaning and referent are 
complex, it is not necessary for each element of meaning to correspond 
to an element of the referent, and vice versa. He quotes the instance 
a land without mountains from Bolzano (Husserl II. i. 296 sqq.). I am 
not certain that this is correct; negative expressions stand in a class 
of their own, and require special investigation. 

Phrase meaning is, consequently, not to be interpreted as necessarily 
corresponding to a complex referent; it may owe its complexity to 
the fact that the subjective apprehension of a simple referent is com- 
plex. On the other hand, in a majority of cases the single words in a 
sentence probably correspond to details and relations within the total 
referent. At any rate, we have to assume that each word in a sen- 
tence carries its own meaning as contribution to the phrase meaning, 
since the latter is usually ruined if one word is removed. Some of 
these word-meanings are basic, others relational (cf. 5.51). 

In simple cases, we immediately understand the whole phrase by 
understanding the single words. In other cases, we may be famihar 
with every word used, and still fail to understand the meaning of the 
whole (see 6.38). The comprehension of a sentence may involve in- 
tellectual operations of considerable difficulty. We do not understand 
a sentence if we do not understand the single words composing it. If 
a German is told that how do you do? means Guten Tag, he still does 
not really understand the English phrase. He knows its referent, 
but not the precise meaning (Ammann, Rede I 44 — 45). 

Stahlin has attempted to disentangle word-meaning from phrase- 
meaning in some of his experiments, but with a negative result: he 
arrives at the conclusion that it is useless to ask for the meaning of 
the single words in the phrase, because there exists no apprehension 
of the word-meanings distinguishable from that of the total phrase- 
meaning. Our apprehension of the whole is disturbed if any single 


word becomes too obtrusive. The study of single words is artificial 
(Stahlin, Exp, Unt. 155—157). 

This statement cannot be accepted. There is no getting away from 
the fact that single words have more or less permanent meanings, that 
they actually do refer to certain referents, and not to others, and that 
this characteristic is the indispensable basis of all communication. (Cf. 
3.25 on the traditional range). It is on this basis that the speaker 
selects his words, and the hearer understands them. It is true that 
in the case of familiar phrases we may treat the whole phrase as a 
unit of speech and comprehension, referring it to the total referent 
without further analysis. But the removal of any single word would 
disturb the correlation. And simple phrases do not constitute the 
whole of language. That Stahlin has not been able to find the word- 
meanings in the phrase-meaning is no proof that they are not there. 
His not finding them is easily explicable: the word-meanings are not 
the final end of speech, but merely a means to that end; they are 
passed over and disappear as soon as they have done their work. Their 
intensity may, as Spearman has shown (cf. 4.17) be very low, so that 
they are difficult to introspect, but their determinateness may still 
be sufficient for the purpose of building up the phrase-meaning. The 
highly automatized speech function is capable of going via the word- 
meanings to the phrase-meaning, if no obstacle presents itself, in pract- 
ically no time. Biihler's experiments show clearly enough that if 
sufficiently difficult phrases are given to the observers, they have to 
work on the basis of the word-meanings in order to arrive at a com- 
prehension of the phrase. 

The study of single words and their meanings is the indispensable 
basis of scientific semasiology.^) 

4.38. Autosemantic and Synsemantic Meanings. Marty, and con- 
sequently also Funke, have claimed a fundamental importance for the 
distinction of autosemantic and synsemantic expressions. (See F'unke, 

^) It seems uncertain whether relational meanings belong to word meanings or 
to phrase-meaning, or if both are possible. In the sentence previously quoted. 
Jack beats Jill, the word-order stamps Jack is subject and agent. Are we to say- 
that the relational element thereby connected with the meaning of Jack belongs 
to the meaning of the word, or forms an element of the phrase-meaning? The 
latter view is perhaps more consistent. But compare a Latin sentence: Leo cervum 
inte r fecit. Leo is shown by its form to be the subject, and the word-order is of 


Innere Sprachform 22 sqq.). There are, lie says, in language " Ans- 
el rucksmittel, die schon fiir sich allein genommen den Ausdruck eines 
fiir sich mitteilbaren psychischen Phanomens bilden, wahrend es da- 
neben andere Sprachmittel gibt, von denen dies nicht gilt". In actual 
speech only so-called sentences are really auto-semantic. "Sie bilden, 
indem sie im Horer Urteils- oder Interesse-phanomene zu wecken be- 
stimmt sind, die wichtigsten, fiir sich verstandlichen und abgeschlos- 
senen Gedankenstiicke der Rede". Autosemantic are further Vor- 
stellungssuggestive, which are intended to evoke Vorstellungen (images?) 
in the hearer's mind. These are divided into fingierte Reden and A''a- 
men. The former are for instance poetical assertions, exclamations, 
etc. The latter, "bedeuten begriffliche Vorstellungen, wie z. B. Hans, 
Garten, Ding, Berg, etc." They are not used alone in actual speech, 
"stellen aber doch hinsichtlich ihrer Bedeutung unter dem Vorrat an 
Ausdrucksmitteln gewisse relativ selbstandige Zentren dar". 

The synsemantic words, on the other hand, are "alle die, welche nur 
mit anderen Redebestandteilen zusammen eine vollstandige Bedeutung 
haben, sei es, dass sie einen Begriff erwecken helfen, also bloss Telle 
eines Namens sind, oder zum Ausdruck eines Urteils (einer Aussage) 
oder zur Kundgabe einer Gemiitsbewegung oder eines WiUens (zu einer 
Bitt-, Befehls-formel, u. dgl.) beitragen." Such are the prepositions, 
conjunctions, adjectives, certain verb forms, (as goes, stands, infinitives 
and participles), subordinate clauses (as who has beaten; that he was ill), 
casus obliqui of the autosemantic nouns {the father's), elements of com- 
pounds {church-tower , as compared with church and tower, which are 
autosemantic) . 

As Husserl points out (II. i. 307) we have to distinguish not only 
between autosemantic and synsemantic expressions, but also between 
autosemantic and synsemantic meanings. With regard to the syn- 
semantic word, whenever it functions in a normal way, in actual speech, 
it carries a certain dependent element of meaning and so gives its 
contribution to the whole (1. c. 306 — 307). The dependency of the syn- 

no importance in that respect, cervum leo interfecit would be equally correct and 
clear. Are we to assume that in a synthetic language the relational elements 
entering into the meanings of phrases are apprehended as adhering primarily to 
the word-meanings? While in an analytic language the sentences are apprehended 
as complexes into which the relational meanings enter as factors of equal value, 
not as appendages of the basic meanings? 


semantic meanings is not to be explained as due to their being appre- 
hensions of dependent referents. The expression dependent referent 
itself is a sufficient refutation. It is an autosemantic expression, but 
it denotes something dependent. Any dependent element of a referent 
may thus be made the referent of an autosemantic expression: redness, 
similarity, size, unity, existence; and this applies both to independent 
elements of basic referents, and to relations and categories (1. c. 313). 

The referents of isolated synsemantic expressions exist only as 
qualities of other referents, or in relation to other referents, and so 
on (cf. above 4.262). The meaning of the isolated word and is 
realized by us either through the vague thought that this is a certain 
well-known particle, or else by the help of vague images or thoughts 
constituting a combination of the t3^e A and B (1. c. 316). 

It seems evident that if an expression denotes a referent that is (i) 
a characteristic of another referent {green hat), or (2) a relation be- 
tween other referents (black or white), or (3) placed in a definite rela- 
tion {father's), or (4) an action of another referent (the arrow flies), 
and so forth, then we require some supplementation for complete un- 
derstanding. This supplementation is not necessarily given in words; 
it may be given in a perceived or otherwise apprehended context. 

In my study of the sense-developments of a group of adjectives and 
adverbs, I found that some changes of these words were conditioned 
by the nature of the governing word, or properly speaking, by the na- 
ture of the referents denoted by the governing words. Thus the at- 
tribution to living beings influenced the meanings of adjectives in a 
certain way (Stern, Swift 216). Such factors will have to be taken 
into consideration in analysing sense-changes. At present, for a de- 
scriptive analysis, it is not necessary to go into the matter in detail, 
since it belongs to phrase-meaning rather than to word-meaning. 

4.3- A Note on Other Definitions. If the argumentation in the preceding chap- 
ters is convincing, other theories of meaning will have to be rejected. However, 
a few words may usefully be added with regard to some of them. 

Hermann Paul (Prinz. 75) distinguishes okkasionelle und usuelle Bedeutung, 
giving one definition for each: "Wir verstehen also unter usueller Bedeutimg 
den gesamten Vorstellungsinhalt, der sich fiir den Angehorigen einer Sprach- 
genossenschaft mit dem Worte verbindet, unter okkasioneller Bedeutung den- 
jenigen Vorstellungsinhalt, welchen der Redende, indem er das Wort ausspricht, 
damit verbindet und von dem er erwartet, dass ihn auch der Horende damit ver- 


The formulation seems to imply that each word has one usual meaning only, 
a synthesis of all its occasional meanings (similarly 1. c. 82 and 84). But in 
other places (1. c. 77 — 78) Paul expressly speaks of a word having more than 
one usual meaning. Taking the definition as it stands, usual meaning is approx- 
imately equivalent to what I have called lexical meaning (4.22), but it is 
confused with general meaning and with the range of the word. According to 
Marbe and Marty, Paul also employs the term in the sense of habitual meaning. 
A term of so many interpretations is clearly useless. 

With regard to occasional meaning, Paul provides an ingenious differentia 
specifica to his genus proximum, Vorstellungsinhalt. It is based on the circum- 
stance that out of the many possible particular meanings within the range of a 
word, the speaker intends only one to be apprehended by the hearer, as for in- 
stance in give me a crown (= five-shilling piece), where all other alternatives are 
excluded by the formulation. But, as I have shown in 3.16 above, the formula- 
tion is much too wide, because it does not exclude various items of context. When 
Caesar reported to the Senate, "Veni, vidi, vici" , he intended the Senate to under- 
stand that the campaign was over and the enemy conquered — but that is an 
inference, not the meaning of the phrase. 

Paul's evident belief that a word, in occasional use, had one meaning, common 
to speaker and hearer, is an untenable position (cf. 6.5). Another weak point 
is his assumption that the connection between a word and its meaning is only 
an association, a view rejected by Wundt and all later authorities. (Cf. further 
Stout I 78 sqq., II 216 sqq., Marbe 493 sqq.; Weisgerber, GRM 15, 169; Ahlmann 
14 sqq., and 3.21 above). 

E. Wellander improves on Paul in one way, by distinguishing the meanings 
attached to the word by speaker and hearer, and attempts to comprise what he 
calls individuelle und lexikalische Bedeutung in one definition: "Die Bedeutung 
eines Wortes ist die Vorstellung die ein Individuum mit diesem Worte verbindet" 
(Studien I 7). We are not told what Vorstellung is; if we take it to mean mental 
content, which seems a likely interpretation, the definition is too wide to be of 
any use (cf. Stern, Litteris III 50 sqq.). Wellander defines individual mean- 
ing as "die Bedeutung die das Wort in dem jeweiligen Zusammenhange hat", and 
also as "die tatsachliche Bedeutung eines Wortindividuums" (1. c. 12 and 17). 
Bedeutung is here used as genus proximum, and if we insert the definition of that 
term in the last two formulations — which ought to be possible — they state, 
one, that individual meaning is the mental content {V orstellung\) which a speaking 
individual connects with the word in a specific context; the other, that individual 
meaning is the mental content actually connected with a word in any individual 
case. It is evident that such vague formulations are of very little use; they are 
scarcely more than other names for the definiendum. Moreover, they build on 
the view that "Wortvorstellung" and "Sachvorstellung" are connected merely 
by an association, and thus, in a psychological and epistemological question, 
completely neglect the unanimous opinion of experts on these matters. 

H. Gomperz (II. i. 69) describes meaning as a relation. He argues in the fol- 
lowing manner. What is it that means? It cannot be the word alone, for that 


is merely a collocation of soimds. A psychic element must be added to it if it 
is to mean anything, and that is the content [Aussageinhalt). It is a complex 
of word and content that has meaning. Gomperz calls this complex sinnvolle 
Rede or Aussage. 

What is meant by the Aussage? It cannot be merely the referent {Aussage- 
grundlage) as such, for we have seen that a referent may be apprehended in many 
different ways (cf. 3.24). What is meant is the Aussagegrundlage, apprehended 
in a certain way, and this apprehension takes place through the content [Aus- 
sageinhalt). What is meant is therefore a complex of referent and content, which 
Gomperz calls der ausgesagte Sachverhalt. The relation between Aussage and 
ausgesagter Sachverhalt — into both of which the Aussageinhalt in a peculiar 
way enters — is Bedeutung (a term used only of actual meaning). 

It is evident that Gomperz takes the problem from a logical point ot view, 
while for my purposes I have to take the empirical psychological position. Never- 
theless, Gomperz' masterly analysis of the interrelations of the three factors, 
word, meaning, and referent, is of the greatest importance, and his is the first 
tenable definition of meaning. It is regrettable that it has been so little noticed. 

For further details on this matter I refer to Ogden-Richards (305 sqq.), who 
give a list of sixteen different types of definitions, with comments on them. 


5.1. Introductory Remarks. When the psychic processes connected 
with the production and comprehension of speech run their habitual 
course, the meanings of words remain stable. If, for some reason, a 
modification occurs in the habitual succession of mental events, a modi- 
fication of meaning may ensue. In order to show, if possible, at what 
point in the process the changes set in, what items are modified, and for 
what reasons, it is necessary to compare the changes with the "normal" 
state of things. I shaU therefore give, in this chapter and the following, 
a summary of present opinion concerning the processes involved in the 
production and comprehension of speech. 

With regard to the production of speech, I shall try especially to as- 
certain at what point the word arises in the speaker's mind. If this 
occurrence takes place at a definite stage in the development of a 
thought intended for expression, it might be possible to gain some 
further data for the explanation of sense-changes. We should, so to 
speak, be able to ascertain the precise psychic situation in which the 
word arises, and thus perhaps to infer the influences to which its 
selection is subjected. 

In normal speech, the words arise automatically in response to the 
impulse to say this or that, or perhaps merely the impulse to say some- 
thing. The whole process is so instantaneous that it evades analysis. 
Similarly, comprehension of speech is normally an automatic process. 
A possibility of analysing the process is afforded in two ways; in aphasia, 
where the speech functions are more or less broken up; and by artificially 
retarding the speech process, for experimental purposes. 

I shall first give a summary of some facts ascertained in modern 
research on aphasia, as far as they are of interest for the present work. 

5.3. Research on Aphasia. 
5.21. Preliminary Remarks. In the following account of aphasic 
defects I make use of H. Head's "Aphasia and Kindred Disorders of 


Speech", the most recent authoritative statement on these matters, 
and one that pays much attention to points that are of special interest 
to linguistics. The main points of Head's theories are to be found also 
in three papers in Brain and Brit. J. of Psych. (Cf. Cassirer, Phil. Ill 
241 sqq). 

Aphasia is, according to Head, a failure of the capacity for symbolic 
formulation and expression, a capacity that is indispensable for the per- 
fect use of language. Symbolic formulation and expression "is a mode 
of behaviour in which some verbal or other symbol plays a part between 
the initiation and execution of the act". It is a purely descriptive term, 
and the processes to be included in it must be decided empirically (Head 
I 211, 399, 423). 

The manner in which symbolic formulation and expression inter- 
vene in very simple acts is illustrated by comparison with acts of direct 
reference. "Suppose some article such as a knife, or even a geometrical 
figure, is placed in his hand out of sight, the patient has no difficulty 
in selecting its duplicate, provided that no words are employed. More- 
over, if he has been given a pyramid cut out of a block of wood, he can 
match it with any pyramidal object, however greatly the two may 
differ in relative size and structure. He deduces from the multifarious 
sensations yielded by his hand certain characteristics, which are also 
possessed by the object within sight, and ignores the many differences. 
In both percepts he reacts to a common quality, the pyramidal factor, 
although the one is the result of tactile, the other of visual impressions. 

So long as the act to be performed is one of direct matching it can 
usually be excuted in spite of the defective use of language. But as soon 
as a symbol intervenes between the initiation and performance of an}- 
task, the patient is liable to fail to carry it out correctly. Suppose, for 
instance, he has succeeded in matching a single object shown to him 
with the one on the table which resembles it. If he is then given two 
objects at a time, he may fail to select the two corresponding duplicates , 
because he attempts to register what he has seen in words and to 
make his choice accordingly. A symbolic formula has been interjected 
and the act is no longer one of direct matching" (Head I 517 — 518; 
212, 385 sqq.; cf. also below 5.25). 

It is the function of words as symbols that is disturbed by aphasia. 
I have distinguished above (2.22) between symbols and signals. 
We find this distinction corroborated here. "There are certain acts of 


speech which have little or nothing to do with thinking, and neither state 
a proposition nor culminate in action. These remain unaffected in 
aphasia and kindred disorders. They comprise meaningless words and 
phrases, emotional ejaculations, such as "Oh! dear me", together with 
oaths and other familiar expletives. . . 

Even in current speech there are many words which, though used 
idiomatically, have little distinctive meaning in themselves. Many 
ready-made expressions or cUches are employed habitually in order to 
start and maintain the progressive flow of speech. These are Unguistic 
tricks, which enable the speaker to utter the essential contents of his 
mind, and many of them give emotional tone to an otherwise colour- 
less phrase. I once took down the following sentence from the lips of 
an irate colleague: "(I'll tell you what it is,) he (jolly well) wants (to get) 
a (common or garden) secretary to make him answer his letters". All 
the words in brackets are logically unnecessary, but they helped him to 
express both his opinion and his irritation. 

Slang closely approaches this order of verbal utterance. Many edu- 
cated aphasics, when in difficulty, fall back on this more descriptive and 
ready-made method of expression" (Head I 516). The contrast here 
discussed comes out very clearly in comparing two kinds of automatic 
expressions or phrases constantly employed in everyday speech. "The 
one consists of ejacvdations and phrases devoid of logical meaning, which 
serve to betray emotion or to form the preliminary to significant verbal- 
isation. These escape altogether in aphasia, for they have little to do 
with systematic thinking. On the other hand, there are many acts of 
speaking and understanding spoken words, which, although they have 
become by practice almost habitual, remain endowed with significance. 
However great the facility of diction or of comprehension, these proces- 
ses were developed out of formal thinking and still serve to secure that 
end. They consequently suffer severely in disorders of symbolic form- 
ulation and expression" (Head I 516; cf. 142, 385). 

In a similar way visual images are affected in so far as they have sym- 
bolic value for thinking; see the quotation in 4.142 above. 

Even if, primarily, the aphasic patient's intelligence is unimpaired, 
his defective power of manipulating language will not only make him 
appear stupid, but will throw him back on more primitive methods of 
thinking and acting. "If I am told to take the second turning to the 
right, I am precluded from choosing any one on my left hand and also the 


first on the right. But many aphasics, unable to comprehend or to 
retain the exact terms of a command, fall back on the method of trial 
and error. Having taken a false turning, they look around to discover 
that the objects actually in sight do not correspond with those they ex- 
pected. They then cast back and explore other ways until, catching 
sight of some familiar landmark, they walk on confidently towards 
their goal" (1. c. 524). The severer forms of aphasia involve still greater 
disabilities, since logical thought is in the main impossible without the 
help of symbolic formulation and expression. The importance of lan- 
guage for thought comes out convincingly in such cases. "The lesion does 
not of necessity destroy the power to think, but one method by which 
thought is carried into action is disturbed. The patient is robbed of 
certain forms of effective symbohc representation and has lost the 
normal means of communicating with his fellows" (1. c. 394; cf. 418). 

Head severely criticises the opinion that there are in the brain "speech 
centres" (cf. Froschels 19 — 20, with various quotations, and Pick, 
Sprachstor. 21 sqq). It is true that a lesion of certain specified parts of the 
brain will lead to disorders of speech, but this shows only that such 
parts of the brain are "nodal foci, where central neural activities 
undergo integration and other changes in relation to one another" (Head, 
I 474). "An act of speech is a march of events, where one chang- 
ing condition passes insensibly into another. When speech is defect- 
ive, this easy motion or transition is impeded, one state cannot flow into 
another because of some mechanical imperfection in the process . . . The 
site of such a breach of continuit}^ is not a "centre for speech", but 
solely a place where it can be interrupted or changed" (1. c). 

5.22. The Classification of Speech Defects. Earlier theorists have 
distinguished motor and sensory aphasia, the former being inability 
to produce speech, the latter inability to comprehend it. Another 
theory distinguished defects of speaking, writing, or reading. These 
theories have been criticized by several writers (cf. Pick, Sprachver- 
standnis), and Head rejects them as not compatible with the clinical 
facts (Head, Aphasia I 134 sqq). 

Head's own point of view is that "the various disorders of speech pro- 
duced by injuries of the brain manifest the ways in which the organism 
masters a situation, demanding the use of language, with a defective 
mechanism. A certain form of behaviour becomes necessary as a sequel 
to certain external or internal events; some normal facility is disturbed 


by the presence of the lesion and the orderly exercise of a series of func- 
tions suffers in consequence. A new attitude must be assumed; for the 
patient has to face a familiar situation with an imperfect apparatus. It 
is as if he were compelled to play lawn-tennis with a broken racquet; 
many of his favourite strokes will become impossible and he will have to 
vary his conduct in accordance with the defective instrument in his 
hand. The movements he adopts in consequence of these unusual con- 
ditions do not form integral parts of his normal method of playing the 
game. A man who has a pain in his toe walks differently from one whose 
heel is affected. But neither gait reveals the elements out of which norm- 
al walking is composed. Both are due to the assumption by the indi- 
vidual of a new functional attitude in face of abnormal conditions" (Head 
I 301). Similarly in aphasic disorders, the undisturbed functions do not 
go on as under normal circumstances, but adapt themselves to the ab- 
normal situation, attempting to make up for the deficiency. Each case 
of aphasia represents the response of an individual patient to an ab- 
normal situation (1. c. 428). 

With due regard to these and several other points, which cannot here 
be detailed, Head considers that "we are justified in recognising the 
existence of certain classes of aphasia. For the clinical manifestations 
are so obviously different according as the loss falls mainly on one or 
other group of functions necessary for language in its widest sense, that 
some formal differentiation or grouping of the phenomena is necessary" 
(1. c. 220). 

I shall give a brief summary of the main characteristics of the four 
classes of defects described by Head, keeping chiefly to those character- 
istics that are of interest to the present investigation. The overwhelm- 
ing mass of details in Head's book makes a summary difficult, but I 
have attempted to bring out the most important points. 

5.23. (i) Verbal Defects. "Any disturbance of this aspect of sym- 
bohc thinking and expression is revealed primarily by defective word- 
formation. The patient is unable to find the words he requires for 
ordinary conversation; in the severest cases he may be reduced to"yes" 
and "no", together with a few expressions, which he employs automat- 
ically or solely under the influence of emotion. So grave a disorder of 
articulated speech is always accompanied by some loss of power in 
writing and want of verbal memory for the content of sentences read 
silently. . . However fluent they may ultimately become, these patients 


always find difficulty in pronunciation. Words of more than one syl- 
lable tend to be slurred and shortened. No. 4 complained that he had 
difficulty with "tenical terms" (technical terms). "Yesterday", he said, 
"I had diff-ulty in remembering what you do with skull . . . tri . . . tre. . . 
trephine ..." No. 6 spoke of "the claration of war bytheOUies" (declar- 
ation of war by the Allies). Moreover, especially in rapid conversation, 
words were dropped out in the struggle to convey the desired meaning; 
but there was none of that omission of the syntactical parts of speech 
which in other forms of aphasia leads to a "telegraphic style" of utter- 
ance. (Head I 221 — 228). 

"These patients can draw, play card games and enjoy jokes set out 
in print or pictures. In fact, the disorder from which they suffer affects 
mainly verbal structure and words as integral parts of a phrase; their 
nominal value and significance are perfect, except for the disturbance 
produced by articulatory abnormalities which affect both external and 
internal speech" (1. c. 413). 

Valkenburg states concerning one of his patients: "Mit der gehorten 
Rede wusste sie anfangs nichts anzuf angen, sie sah den Mund des Sprech- 
enden sich offnen und horte die Klange, die ihr keine Worte waren" 
(Valkenburg 3). This was clearly a severe case, involving complete in- 
ability to combine the sounds into words. Further instances, see Pick, 
Sprachverst. 66, 68. 

5,24. (2) Syntactical Defects. "This form of disturbance of language 
can be distinguished by the fact that the patient talks jargon. In other 
varieties of aphasia he may not be able to evoke the word he desires 
to use, and, in his efforts to find it, gives vent to sounds that do not 
correspond to any recognisable language symbols. If he is of a lively 
temperament he tries again and again to correct his faiilty nomencla- 
ture and may fly to metaphorical expressions in order to circumvent 
his want of ability to express his meaning; but this cannot be 
described as true jargon. 

On the other hand with a syntactical disorder of language the pa- 
tient talks with great rapidity, when once started. Individual words 
may be recognisable, but the grammatical structure of the phrase is 
liable to be badly affected. He talks fluently in short jerky sentences, 
slurring or omitting many of the junction words. Even when they are 
present, it is difficult to hear the articles, conjunctions and other com- 
ponents necessary to a perfectly formed sentence. Asked what he had 


done since his admission into the lyondon Hospital, No. 15 said, "To 
here, only washing, cups and plates. That's about all you've got to 
do here". "Have you played no games?" I enquired, and he repUed, 
"Played game, yes, played one, daytime, garden". . . . 

"Not only are the rhythmic movements of the phrase affected, but 
the internal balance of its constituent elements is disturbed. The pa- 
tient cannot "touch off" the words so as to produce an accurately 
coherent sentence, and the artictilatory rhythm of polysyllables tends 
to be disturbed". (Head I 230). "The phrase may be fatdty and 
even single words are badly pronoimced, although their nominal use 
and meaning remain intact. All those smaller words, such as articles 
and conjuntions, which bind together the more significant parts of the 
sentence, tend to be slurred or dropped. Speech becomes a series of 
disconnected categorical statements" (1. c. 239). 

"These patients can understand the full meaning of picttues but 
they are greatly hampered by their jargon, if they attempt to convey 
to others or silently to themselves what they have gathered . . . This 
disorder is essentially one of balance and rhythm in symbolic expression, 
and syntax suffers greatly. The patient has plenty of words, but their 
production is ataxic and they are strung together without the usual 
connecting links. This leads to jargon and renders difficult even in- 
ternal formulation of words and their meaning" (1. c. 414. Cf. Val- 
kenburg 17 — 18, Isserlin 360 note, with a typical instance of "Agram- 
matismus"; Froschels 84 sqq.; Ivaguna 245, who points out that the 
successful use of language depends on the abiHty to discriminate ob- 
jects in their relations to one another, independently of their direct 
relation to the person who is speaking). 

5.25. (j). Nominal Defects. "In this disorder of symboUc formu- 
lation and expression, we are not deaUng with a difficulty in shaping 
words or phrases, but with a disturbance of their nominal significance. 
A name is a pattern which, if appropriately chosen, fits an external 
object or state of things around us. . . So far as (the patient) can 
find words they are enunciated correctly and united into coherent 
phrases. There is none of that profound disturbance of pronunciation 
which runs through aU the utterances of the verbal aphasic; nor are 
the sentences jerky and ill balanced, as with the syntactical defects 
of language". (Head I 240 — 241). 

"During the tests with colours. No. 2 made such gross mistakes that 


he might have been thought to be colour-bHnd; for white was called 
"green", black "red", and green "blue". Exactly the same kind of 
error occurred when he chose a colour from its printed name on a 
card; on this occasion he even chose white for black, and black for 
green. More or less similar mistakes were made by all the patients 
of this group, and No. 22 was sent to me as an example of aphasia 
with colour-blindness. But not one of them had the sHghtest difficulty 
in choosing, from amongst the colours on the table, that which matched 
the one I had shown him. From these observations we might be 
tempted to think that they had lost this knowledge of the nature 
of colour, that they were in fact "mind-bHnd". But No. 2, in his 
attempts to explain to me his difficulty in reading the printed cards, 
began to point to my white coat, to his khaki tie, the blue band on 
his arm which he wore as a wounded officer, and the green of the 
trees outside his window. Instead of the names of the colours, he 
was therefore encouraged to use a set of similitudes; black was "what 
you do for the dead"; red "what the Staff wear", or, pointing to the 
lapel of his tunic, "where the Staff have it", and so on" (1. c. 242).^) 

"Nearly two years after the injury No. 2 was able to fill up a cheque 
spontaneously; but he could not be certain that the written words and 
figures corresponded with one another. He drew a cheque in my pre- 
sence for eighty-five pounds, ten shillings and sixpence, but filled in 
the figures as £80.10.6. He noticed this discrepancy and succeeded 
finally in making the correction; but the cheque would not have passed 
the bank" (1. c. 253). 

"(No. 2) was above the normal average at chess, but could no longer 
play bridge; "The names of the cards bother me", he said. "It's 
just names; I used to play a good game at bridge" . . . These defects 
of speech consist essentially of loss of power to employ names together 
with want of comprehension of the nominal value of words and other 
symbols. Although the patient has plenty of words at his command, he 
may be unable to designate famihar objects; yet he can describe their use 
or composition, either directly or in some apt metaphorical phrase, 
and he can repeat anything said to him provided it is simple and easy 
to understand. Asked to point to an object named by the observer, 
he is unable to do so or makes his choice slowly and with effort. He 

^) A similar case from Gelb and Goldstein, Uber Farbennamenamnesie (Psych. 
Forsch. VI, 127 sqq.) is quoted by Weisgerber, GRM 14, 242. 

Goteb. Hogsk. Arsskr. XXXVIII: i 


reads and writes with extreme difficulty" (1. c. 256 — 257, 414; Valken- 
burg 17). 

5.26. {4). Semantic Defects. "These patients tend to talk rapidly 
as if afraid of forgetting what they wanted to sa}^' at times this ac- 
tually occurs and the conversation tails away aimlessly. They suffer 
from no difficulty in pronunciation and, although the sentences may 
be somewhat short and jerky, syntax and intonation remain undis- 
turbed. Pure verbal repetition is in no way affected. They can name 
common objects and indicate without fail the one that has been men- 
tioned orally or in print. . . The fault is essentially a want of recogni- 
tion of relative significance and intention. Everything tends to be 
appreciated in detail, but the general significance is lacking. This is 
evident, when the patient is given a picture and told to say what he 
sees in it. He looks at it like a child, pointing out one thing after 
another, and not uncommonly misses some important feature; asked 
what the picture means, he may be entirely at a loss and either gives 
up altogether or invents some preposterous explanation. 

I showed No. 24, who was an ardent student of poUtics, a cartoon 
of Mr lyloyd George playing the harp from the same score as M. Briand, 
who held in his hand a French horn. After contemplating this pic- 
ture for some time he said, "It's the Welsh Prime Minister with the 
Celtic instrument and the other man has a musical instrument, a blow- 
ing instrument. He's a foreigner probably; whether he's a French 
editor, I don't know". I then uncovered the legend beneath the pic- 
ture, which ran, "The World's Premier Duettists", and he replied, "I 
don't understand it; it doesn't help me" (1. c. 257 — 258). 

These difficulties "are due essentially to want of power to com- 
bine mentally into a single act a series of relevant details . . . mani- 
fested mainly in want of ability to appreciate and retain the ultimate 
significance or intention of words and phrases combined in normal 
sequence" (1. c. 261). 

"In no other form was there the least difficulty in setting the hands 
of one clock in strict conformity with those of another, but patients 
belonging to this group tended to carry out this manoeuvre slowly, 
wi th hesitation, and might even become confused by what is little more 
th an an act of matching. They were puzzled by the significance of the 
ha nds and failed to tmderstand exactly what they were expected to 
do . As a rule they could tell the time correctly and this task was 


performed more easily than any other of this series of tests. But, 
when asked to set the clock in response to either oral or printed com- 
mands, the errors were extremely gross. The long and short hands 
were confused, "to" and "past" were mistaken and even the hour was 
wrongly indicated in some instances" (1. c. 262). 

"These patients cannot play games such as chess, draughts or cards. 
Nor can they put together "jigsaw" puzzles. No. 8 complained, "I 
can see the bits, but I cannot see any relations between the bits; I 
could not get the general idea". This was revealed by his attempts 
to play billiards. He could hit a second ball directly and could poc- 
ket it with ease; but if he attempted to put his ball into the pocket 
off another, he not infrequently struck it on the wrong side. For the 
same reason he was unable to make a simple cannon and was incapable 
of bringing off any stroke from the cushion. He said, "I have so 
much difficulty in thinking out the scheme of it" (1. c. 266 — 267). 

"This form of aphasia is characterised by want of recognition of the 
ultimate significance and intention of words and phrases, apart from 
their direct meaning. But other functions suffer that have no immed- 
iate bearing on verbalisation; for there is loss of power to appreciate 
or to formulate the logical conclusion of a train of thought or action. 

The patient has no difficulty in forming words and can repeat what 
is said to him. But in general conversation his sentences tend to tail 
away aimlessly, as if he had forgotten what he wanted to say. Read- 
ing presents no serious difficulty; but the full meaning is liable to be 
misunderstood. These patients can write, but the results tend to be 
inaccurate and confused; for semantic defects are more liable to disturb 
the connected sequence of what is written rather than its verbal form. . . 
The patient's defects make him useless for any but the simplest employ- 
ment; yet his memory and intelligence may remain on a relatively 
high plane. He has, however, lost the power to co-ordinate details 
into a general formula for internal or external statement" (1. c. 267 

5.27. Remarks on Head's Theory. I have devoted some space to 
this account of Head's theory, owing to its great interest for general 
linguistics, and especially semasiology. The speech activity is clearly 
complex; I have shown above (ch. 2) that it comprises various func- 
tions. It seems reasonable to assume that defects of speech affecting 
single functions might be distinguished. In other words, there should 

100 GUSTAF STERN 5.27 

be a correlation between the results of normal psychology and patho- 
logy. Too much must not be expected in a branch of research that is 
so recent. 

Verbal defects affect the ability to pronounce or comprehend words 
as totals having a meaning, an abiHty that clearly is of basic import- 
ance. Biologically, it presumably developed pari passu with the abil- 
ity to use words as symbols; before that, there would be little occasion 
to work out a system of articulate sound symbols (cf. 2. 22). This 
matter falls outside the scope of my investigation. 

Syntactical defects concern the ability to combine words into senten- 
ces; the capacity of grasping relations is somehow involved, but it is not 
quite clear from Head's account how we shoiild interpret the facts in 
linguistic terms. The disturbance of "balance and rhythm" on which 
Head repeatedly insists, may be explained as due to defective appre- 
hension of relations, since a speaker must grasp the logical interrela- 
tions and relative importance of a sequence of words if he is to know 
which of them to stress, and which to leave unstressed. The ability to 
grasp relations is obviously a basic condition not only of human speech 
but of all higher intellectual development; even anthropoid apes are very 
defective in this respect (see lyaguna 227 sqq., discussing Kohler's re- 
ports). It is not a specific element of the speech activity, and so may 
be left aside here. 

Nominal aphasia involves loss of power to employ names, together 
with a want of comprehension of the nominal value of words and other 
symbols, a defect that apparently affects the mastery of the symbolic 
function: the ability to make a correct objective reference from word 
to referent, or, conversely, to find the way from referent to word. In 
the cases reported by Head the failure to do this was not complete, 
since the patients could make their meaning known in a round-about 
way, by using their stock of unaffected words. This defect resembles 
the Unguistic uncertainty described above (4.17); the patients were 
uncertain about the range of the words. 

The semantic defects apparently concern the abiUty to co-ordinate 
the single meanings and to educe the relations between them (on educ- 
tion, see 4.265 note) in order to build up a coherent whole, and on the 
receptive side, the ability to analyse the speech into its constituent word 
meanings. Further, Head states that this is a "want of recognition of 
the ultimate significance and intention of words and phrases apart from 


their direct meaning" (5.26). It seems evident (cf. 2.13, 3.26, and 
6.38) that the meaning of speech contains supra-summative elements, 
founded on the eduction of relations and correlates. It is also possible 
that such operations are not an adequate explanation of all elements of 
a phrase-meaning, especially with regard to the purposive function. At 
any rate, the understanding of a phrase-meaning presupposes the abil- 
ity, not only to grasp relations between words, but also to educe further 
relations and correlates, and possibly also to grasp additional configura- 
tional qualities, not wholly explicable as the result of eductions. (Cf. 
6.38) . The purposive function of speech is probably based on such supra- 
summative elements of phrase-meaning. Semantic aphasia would 
then be the inabiUty to perform the mental operations involved in de- 
ducing the phrase-meaning from the basic data provided by words and 
context. The instance of the political cartoon quoted in the preceding 
paragraph is a good illustration. 

Head reckons among the symptoms of this type the lack of orienta- 
tion, the inability to set the hands of a clock, to lay the table, or to de- 
scribe a picture with proper stress on its main features. I suggest that 
such symptoms should be explained as due to the inability to see in the 
respective situations the intrinsic pattern or configuration. One must 
be able to grasp the pattern of a picture in order to have one's atten- 
tion carried to the main features, giving them their due prominence, 
and neglecting what is merely the filling in of outlines. I^acking this 
pattern, a picture is a jumble of details, and the patient has no guide in 
discriminating what is important from that which is not. He will be 
liable to miss the essential points, and to dwell on subordinate details 
that happen to catch his eye. Similarly in orientation. In laying the 
table it is necessary to be able to correlate the function of the various 
pieces, and in setting the clock to correlate the function of the hands to 
that of the figures on the face. Whether this process should be described 
as an analysis of relations within the complex total, or otherwise, 
is a problem that I have to leave open. We are dealing here with the 
most obscure phase of the speech process, on which normal psychology 
has as yet very little to tell us. One point that requires elucidation is 
the distinction between syntactical and semantic defects, in terms of 
linguistics and normal psychology. Both types are concerned with re- 
lations. — Some further remarks on these matters will be made in later 

102 GUSTAF STERN 5.27 

Among those who have criticized Head's theories, van Woerkom 
suggests that, in order to construct a perfect phrase, we must be able 
to apprehend correctly relations in space and time. He thinks that in 
some of Head's instances it would be better to explain the disorder as a 
lack of the notions of time and space, rather than as a disorder of 
symbolic thinking.^) The symbol appears at a relatively late stage of 
linguistic development, when the spatial relations are already estabUshed, 
and the sense of rhythm is being formed. These two functions are 
essential to the further development of symbolic thought (Woerkom 
730 sqq.; cf. Isserlin, and Froschels 35 sqq.). 

5.3. The Evidence of Normal Psychology. 

5.31. Sentence Definitions. Paul and Wundt. Opinions concern- 
ing the production of speech have sometimes been expressed in the 
form of a genetic sentence definition. Paul, and many others, considered 
a sentence as a synthetic formation: "Der Satz ist der sprachliche 
Ausdruck, das Symbol, dafiir, dass sich die Verbindung mehrerer Vor- 
stellungen oder Vorstellungsgruppen in der Seele des Sprechenden voll- 
zogen hat, und das Mittel dazu, die namliche Verbindung der namlichen 
Vorstellungen in der Seele des Horenden zu erzeugen" (Paul. Prinz. 

Wundt, on the other hand, regarded speech formulation as an ana- 
l5rtic process: "(Ein Satz ist) der sprachliche Ausdruck fiir die willktir- 
hche GHederung einer Gesamtvorstellung in ihre in logische Beziehungen 
zu einander gesetzten Bestandteile" (Wundt II 248). 

The controversy on this point is well known: it does not concern us 
except in so far as the method of speech formulation is involved. Is 
it synthetic or analytic in character?^) 

^) Cf. on this point Selz I 174: Demgemass erhalten wir dutch die Bildung um- 
fassender Komplexe eine doppelte Ordnung unserer Bewusstseinserlebnisse, bezw. 
ihrer Reproduktionsgrundlagen: i. eine raumlich-zeitliche Ordnung, 2. eine 
Ordnung durch Beziehungsverkniipfung, in welcher die einzelnen Gegenstande 
des Bewusstseins als Glieder vielfach zusammengesetzter Sachverhaltnisse er- 

2) I refer the reader to the following writers who have dealt with this subject 
from various points of view: Siitterlin 144, Delbriick 136, Oertel 280, Biihler, 
Satz 12 sqq., Frobes II 250, Delacroix 207 sqq., Gomperz II. i. 58, 239, Pills- 
bury-Meader 254 sqq., C. Biihler 194 sqq.; Pick, Sprachstor. 130 sqq.; Stern, Kin- 
dersprache 179 sqq.; cf. also Wundt II 252 sqq., and 4. 27 above. 


The problem has been investigated experimentally by Selz. He gave 
his observers a word which they were to define in a sentence, reporting 
immediately on the mental process leading up to the final formulation. 

I shall give a summary of his results (cf. also Willwoll 35 sqq., and Selz 

II 362 sqq. on other investigations). 

5.32. Sentence Formulation according to Selz. Selz found that sen- 
tences may be formulated in more than one way, and he has tabulated 
his results in the following manner. "I. Die Satzreproduktion, die in- 
soweit stattfindet, als der Sprachinhalt schon vor Beginn der Formu- 
lierung feststeht und nur reproduziert zu werden braucht. — This type 
is of no further interest to me. 

II. Die Satzbildung, bei der der Sprachinhalt erst neu zu finden ist. Je 
nachdem auch der Gedankeninhalt erst neu zu finden ist, oder schon 
feststeht, ergeben sich hier zwei Unterfalle. 

A. Der Gedankeninhalt ist schon gegeben und braucht nur zum 
Zwecke der Formulierung auseinandergelegt zu werden. Dies ist 
der Fall der analytischen (nachtrdglichen) Formulierung. 

B. Der Gedankeninhalt ist ebenso wie der Sprachinhalt neu zu fin- 
den. Im Gegensatz zu den beiden bisher erwahnten Aufgaben der 
Satzreproduktion und der Satzbildung bei bereits feststehendem Gedan- 
keninhalt, fanden wir fiir die dritte Aufgabe der Satzbildung bei erst 
zu findendem Gedankeninhalt zwei verschiedene lyosungsformen vor: 

a) die phasenweise Formulierung, bei der Gedankenentstehung und 
sprachliche FormuHerung parallel nebeneinander herlaufen; 

b) die synthetische (nachtrdgliche) Formulierung, bei der die zu- 
sammenhangende FormuHerung erst nach abgeschlossener Gedan- 
kenentstehung bei der Zusammenfassung des sukzessive gefunde- 
nen Gedankeninhalts einsetzt" (Selz II 361). 

Note that all this refers to the mental operations preceding the 
verbalization or accompanying it. The formulation of the lin- 
guistic sentence is, according to Selz, always a synthetic process, a 
process of analogy founded on linguistic schemes (1. c. II 362). 

I shall give some details concerning each of the three types. 

5.33. Analytic (Posterior) Formulation. It will be useful to give 
first a typical instance. 

Stimulus word: Stiftung. Task: Definition. Report:'^) "Wusste von 

^) Passages in brackets are explanations added by the observer in reply to 
questions from the experimenter. — Selz' italics. 


vornherein, dass die Aufgabe nicht schwer fallen wiirde, well mir der 
Begriff der Stiftung nach meiner Meinung vollig prdsent war, so dass ich 
ihn vermeintUch nur in Worten auseinander zu legen brauchte, um eine 
Definition zu finden. Es war sicher die vollige Klarheit des Besitzes 
da, nicht nur die Klarheit, dass ich den Inhalt besitze, sondern auch 
der Inhalt selbst. Ich meine, es liegt alles darin, was spdter in die Defi- 
nition kommt; es hegt in einem Male so darin, ich hrauche es nur heraus- 
zuziehen. Ich fing an, mir den Begriff deutHch zu vergegenwartigen. 
Es lag darin, dass Kapital, Zins und bestimmter Zweck vorhanden sei, 
fiir den die Zinsen aufgebraucht werden. Jetzt habe ich erst versucht, 
auf Grund dieser Elemente eine Formulierung zu btlden. . . "Eine 
Stiftung besteht darin, dass ein Kapital zu einem bestimmten Zweck 
gegeben wird, dessen Zinsen. . . "(Nun wiirde gekommen sein: "zu die- 
sem Zwecke verbraucht werden soUen". Diese Wiederholung sucht die 
Versuchsperson durch das Folgende zu vermeiden:) Da merkte ich, 
dass ich die Zinsen sofort auf den Zweck beziehen miisste. Ich begann 
von neuem: "Stiftung besteht darin, dass ein Kapital gegeben wird, 
dessen Zinsen zu einem bestimmten Zwecke verbraucht werden sollen' ' . 
Ich erinnere mich nicht, jemals eine Definition von Stiftung gehort zu 
haben" (Selz II 311). 

In this form, as in synthetic formulation, the mental content of the 
sentence is already more or less clearly present to the mind of the speaker 
before the verbalization sets in. If the task is a definition, the con- 
tent of the defining sentence is present in the form of a unitary total 
which has to be differentiated into its elements and their relations; 
these elements and relations then require to be denoted by words. 

This form agrees in the main with Wundt's theory, with the reserva- 
tion that the total notion {Gesamtvorstellung) need not necessarily be a 
total in imaginal form {ein anschauliches Ganzes) but may consist of an 
awareness of meaning, comprising all the characteristics of the notion, 
but not represented by "eine einheitliche anschauliche Sachvorstellung" . 

The process of differentiation operates on material already present 
in an un-analysed form, and Selz therefore calls it nachtrdgliche Formu- 
lierung. The verbaUzation is analogous to that in the "gradual" form 
(5.34), especially with regard to the influence of linguistic schemes on 
the order in which the various elements of the mental content are made 
the subject of analysis (Wundt: Apperzepfion) . The linguistic scheme, 
as for instance the scheme of a defining sentence, thus contributes in 


determining the progress of mental analysis. I^anguage is not only a 
means of expressing the result of analysis, but through the determining 
action of the linguistic schemes used, it is also a method for the analysis 
of a thought, a concept, or a sensation. 

As in the gradual tjrpe, it is the most general characteristics of the 
mental content, — those that are first clearly apprehended or else 
known from the task (the instruction), — that make possible the 
employment of a definite hnguistic scheme, determining the use of 
further schemes, and guidixig the process of differentiation. Thus, 
when the task was known to be that of defining a notion, the 
observers were influenced into using the scheme of a defining sentence, 
and when it became apparent that the content of the notion was an 
activity, this led to the use of a specialized scheme: "x besteht darin, 
dass . . .". 

The collaboration of linguistic schemes in the formulation of previ- 
ously known mental content makes the actuaUzation and production 
of words fluent and continuous, and the underl5dng mental processes 
become elusive and difficult to analyse. But this must not lead us 
to suppose, with Wundt, that a ready-made sentence hes under the 
linguistic formulation as a simultaneously present totality, which has 
only to be differentiated into successively apprehended parts. (Wundt 
II 244). As in the gradual type, the sentence originates with the actual- 
izing and filling in of the linguistic schemes. The anal5i;ic process con- 
cerns only the mental content. The verbalization, here as elsewhere, is 
a synthetic process, a process of combination (1. c. 350 — 352).^) 

5.34. Gradual Formulation. I shall first give some instances. — 
Stimulus word: Gewalt. Report: "Gewalt ist die Fdhigkeit, etwas ander- 
es in Schranken zu halten .... Ich jing die Definition immer an mit 
Sdtzen "Gewalt is die Kraft . . .", "Gewalt ist die Macht ..." {Der he- 
stimmte Artikel weist hier schon deutlich auf die Antizipation einer ndheren 
Bestimmung hin). Das schien mir aber nicht angangig, solche Worte 
zu gebrauchen. Dann war so eine Verwirrung da die ganze Zeit, ein Auf- 
und Abgehen und Suchen, bis ich auf einmal fand: "Gewalt ist die 
Fdhigkeit ..." Da schien es mir, als oh ich ein harmloses Wort fur die 
Definition gefunden hdtte. Dann ging ich (zundchst) nicht im Satz weiter, 
sondern suchte jetzt charakteristische Eigenschaften herauszufinden {die in 

^) Successful schemes are more easily reproduced again, Selz II 535; see Biihler 
Arch. 12, 85 sqq. on the importance of the sentence-schemes. 


dem durch den hestimmten Artikel schon vorbereiteten Infinitivsatz formu- 
liert werden sollen). Dann kamen die Worte "herrschen", "in Schranken 
halten"; "bedriicken" war auch da. Dann habe ich den mittleren 
Ausdruck herausgenommen . . . Ich wollte die Definition noch einmal 
leise vorsagen, platzte aber laut heraus: Gewalt ist . . ." (1. c. 317). 

Stimulus word: Hypothek. Report: "Garantie ftir geliehene Kapita- 
lien, die in Grundstiicken und Gebauden besteht . . . Das erste, was 
kam, war das Wort Pfand, einfach das Wort . . . das lehnte ich ab mit 
dem Bewusstsein, dass es etwas anderes ist. Dann kam das Wort Ga- 
rantie. Beides sollte die Funktion der Sicherheitsleistung bezeichnen, 
aber das war nicht ausdriicklich gedacht, es kamen einfach die Worte. 
Ich war sehr froh, wie ich das hatte, weil ich dachte dadurch der Losung 
naher zu kommen. Jetzt dachte ich, wofiir ist es eine Garantie, das 
suchte ich. Ich dachte zuerst an Schulden. Es war das Wort da. Das 
wollte ich aber nicht gebrauchen, warum weiss ich nicht. Dann kamen 
auf einmal die Worte gehehene Kapitalien . . . Entschloss mich "geheh- 
ene Kapitalien" zur Definition zu verwenden. Ich sprach innerhch: 
"Hypothek ist Garantie fiir gehehene Kapitahen", wollte damit auf- 
horen. Im diesem AugenbUck fiel mir ein, dass ich dabei das Wesent- 
Hche der Hypothek vergessen hatte, namhch dass es darauf ankommt, 
dass Immobihen belastet sind. Ich zwangte das in einen Nebensatz 
hinein. Nachdem ich das hatte, versuchte ich das noch in einen Haupt- 
satz hineinzubringen, was mir aber zu kompliziert war; dann habe ich 
reagiert" (1. c. 290). 

A sentence may be produced gradually, in due order, even when the 
content of the sentence as a whole is not known in advance.^) It is 
apparent from the reports, that the gradual formulation may begin at 
a moment when no elements of the definite content of the sentence are 
as yet settled. This is possible owing to the fact that in our awareness 
of purpose [Zielhewusstsein) there are anticipated certain very general 
characteristics of the content about to be evoked, and that these char- 
acteristics are hnked up with {zugeordnet) a definite linguistic scheme. 
This scheme is first actualized, and helps to determine the subsequent 
process. Thus, to the general awareness that the sentence is to contain 
a definition, there corresponds the scheme of an "is"-sentence. To the 
only sHghtly more determinate awareness that it is a definition of an 

^) Jespersen, Grammar III. 10. 53, and Miiller-Freienfels, Einfluss 422 quoting 
H. von Kleist: I'idee vient en parlant. Cf. 5.37 below. 


activity, there corresponds the sentence scheme: "x besteht darin, dass. ." 
This abstract sentence scheme, in being actualized, by its determining 
function performs that task in maintaining the coherence of speech 
which Wundt attributed to the concrete total notion {Gesamtvorstellung) . 

The scheme is not only a syntactical scheme, but also a sense scheme, 
or logical scheme for the sentence. It anticipates the parts of speech 
that are to be selected for its completion, not only with regard to their 
syntactical categories, e. g., as nouns or verbs, but also with regard to 
certain general sense characteristics and logical categories. Thus the 
scheme of a defining "is "-sentence not only anticipates a noun to fol- 
low the copula, but also determines the sense of this noun more or less 
clearly as the designation of a genus, to which belongs the species or 
individual to be defined. 

The progress of the gradual formulation is conditioned by the progress 
of the differentiation of mental content on which the scheme is founded. 
Thus, in defining a notion, the necessity may arise of determining more 
particularly the designation already given. The nature of this special- 
ization is generally more or less definitely settled by the nature of the 
previous definition. Every further development of the thought brings 
as a consequence the initiation of corresponding Unguistic processes of 
formulation. The mere awareness that a further specialization has to 
follow, may bring about the actuaUzation of a linguistic formula in the 
shape, e. g., of a relative clause, corresponding to the general de- 
termination. The awareness that the purpose of an object must be 
explained may lead to the actuahzation of a relative clause of the 
type "which serves as . . .". (Cf. similar observations by Biihler in 
recollection experiments. Arch. 12, 33 — 34). 

The continuation of the gradual formulation does not set in until the 
concrete mental content of the new phase is already settled, when, for 
instance, in addition to the fact that a more particular definition is to 
follow, and the direction in which it will lead, its content also is more 
particularly present in mind. But here, too, we are able to dis- 
tinguish two elements in the mental content of the phase to be expressed: 
i) an awareness of the relation in which the sequel stands to what 
precedes, e. g., that it is a more particular attribute, or more 
speciaUzed, or an attribute indicating the end to be served by the 
object previously mentioned, 2) an awareness of the objects that stand 
in this relation. 


The first factor leads to the employment of a linguistic scheme suitable 
for the expression of such relations, for instance, a relative clause. The 
second factor leads to the word for each referent being at once, on the 
actualization of the linguistic scheme, placed in its right position, so that 
for instance the scheme of a relative clause is already complete on coming 
to mind. Especially remarkable in this connection is the high degree of 
abstraction peculiar to the mental content with which the lingmstic 
schemes are linked up izugeordnet) . In one case the scheme of a relative 
clause is reported as being evoked by the general intention to qualify 
the preceding noun more particularly in some way. 

In gradual formulation, the expressions already determined condition 
the selection of words for the sequel. Thus, if a definition is started 
with a noun, denoting the genus, the particular determination to follow 
can be expressed only by a phrase capable of being added to a noun, a 
relative clause or a prepositional phrase. It is no longer possible to use 
an attribute preceding the noun, or a similarly placed prepositional 
phrase, which can always be done in the case of posterior formulation. 
The latter has therefore the advantage of permitting a free choice in 
the arrangement of words. 

The linguistic operations which give to the sentence its logical-syn- 
tactical form correspond neither to a general Regelbewusstsein nor to 
concrete imagery, but to the often very abstract determinations of men- 
tal content and of the preceding and following linguistic content which 
we have called its logical- syntactical structure. This double depend- 
ence of the linguistic operations on the logical structure of the men- 
tal content and on the syntactical structure of the preceding linguistic 
content, shows not only the extraordinary abstractness but also the 
complexity of these correspondencies {Zuordnungen). Concerning their 
origin during the acquisition of language, we are quite as ignorant as we 
are about the correspondence between a linguistic expression and 
the corresponding combination of mental content in the comprehension 
of speech. We may safely assume that at the first actualization of the 
various linguistic operations an important part is played by the ana- 
logical re-construction of experienced combinations between, on the one 
hand, a certain logical structure of the sentence, and, on the other 
hand, the employment of a determined linguistic means of expression 
(Selz II 339—349). 

5.35. Synthetic [Posterior) Formulation. The following is a t)^ical 


instance. Stimulus word: Krieg. Report: "Krieg ist ein Austrag eines 
Volkerstreits mit Waff en . . . Ich habe es gelesen. Zustand der Sicher- 
heit; ich kann das machen. Dann kam mir als erstes das Wort Kampf, 
und weiter: nicht jeder Kampf, das wurde aber nicht gesprochen. Da- 
fiir war ein ganz fliichtiges, unbestandiges Bezogensein auf zwei kamp- 
fende Menschen massgebend. Ich betone, dass das nicht so sein soil, 
also nicht zwei einzelne. Dann kam das Wort "Volk", ich wusste aber 
schon vorher genau durch den Gegensatz des Nichteinzelnen, was ich 
meinte. Das Wort kam als Bezeichnung fiir dieses Gemeinte. Bei 
"Volk" (ausser dem Wort) nichts Sinnliches. Darauf das Moment des 
Gegeneinander, das vorher schon in dem Sinnlichen lag. Ich nam es 
wieder auf und verallgemeinerte es, und zur Bezeichnung fand ich das 
Wort Streit, natiirlich nunmehr auf das Volk bezogen (statt des speziel- 
leren Kampf. Wie das Folgende zeigt, ist Volkerstreit die geeignetste 
nachsthohere Gattung), und nun musste ich dieses AUgemeine (den 
schon als Streit zweier Volker naher bestimmten Streit) wieder spezi- 
alisieren: Nicht etwa diplomatischer Austrag, sondern ein anderer 
Austrag des Streites. Auf diese Spezialisierung war ich gerichtet, ohne 
dass es so ausgedacht wurde. Vielleicht kam mir darauf das Wort 
Waff en. Dann habe ich es zusammengefasst. Kurzer Rtickblick, ich 
sagte dann ja, fing an zu sprechen und wusste, es wiirde richtig heraus- 
kommen, z. B. ergab sich dann das Wort "Austrag", und "Volkerstreit" 
(die Zusammensetzung) ganz von selbst. Diese Worte waren vorher 
nicht da" (1. c. II 296). 

The anal3rtic formulation takes place with the help of an analytic dif- 
ferentiation of previously undifferentiated mental content. Synthetic 
formulation, on the contrary, consists in the combination of elements 
of content which arise one by one and at first are distinct from each other. 
Thus, for instance, the various elements required for a definition are 
found singly, and afterwards combined into a defining sentence. Such 
elements do not arise entirely unconnected, but each added characteris- 
tic appears as a further determination of a previous one, leading in a 
definite direction, and the process is continued until we are aware that 
the materials for the definition are complete, when the ultimate syn- 
thesis takes place (cf. Biihler, Arch. 12, 89 — 90, and Pick, Sprachstor. 
210 sqq.). Paul's definition of a sentence as the expression of the fact 
"dass sich die Verbindung mehrerer Vorstellungen oder Vorstellungsgrup- 
pen in der Seele voUzogen hat", thus fits this t5rpe, but not the other two. 


As already pointed out, the linguistic formulation, the verbahzation, 
is, in contrast to the mental process, always of a synthetic nature. The 
former always consists in the actualizing of hnguistic schemes, and 
the filling of them with new concrete elements. The synthetic nature 
of the linguistic formulation is especially evident in cases where the 
various characteristics of the notion, as soon as they are found, are 
fixed by means of words, which thus form a kind of framework for the 
sentence. The words are at first syntactically unconnected, and are 
combined into a sentence, by the insertion of words denoting the rela- 
tions between their referents. 

The entire structure underlying the mental content recalls the para- 
taxes of child language or primitive speech, which are characterized 
by the lack of designations for the relations between referents (cf. 
aphasia!). In both cases the indispensable elements of characterization 
and linguistic expression are included, while the relations between them 
may be supplied in thought by the hearer as well as by the speaker. 
In the case of sentence formation their designations may therefore be 
left until the stage of verbalization. The expediency of this abrupt inner 
speech, which sometimes also accompanies the gradual type of formula- 
tion, is a sign that we have here to do with determined processes which 
are of value for the fixation of thought, as well as for the preparation of 
the connected formulation, and thus are to be regarded as methods of 

The synthetic type of sentence formulation is a kind of inversion of 
the gradual type, in which the sentence scheme was actualized before 
any of the concrete elements to fill it were yet present in mind. In the 
synthetic type these concrete elements are first designated, and subse- 
quently, in the linguistic formulation, the verbalization, inserted into 
the appropriate linguistic scheme (as Streit, Volk, Waffen, in the in- 
stance quoted). In this case, as in the gradual type, we have to assume 
that linguistic schemes sometimes collaborate from the beginning in the 
constructive process, determining the order in which the single character- 
istics arise in the mind and receive their names (1. c. II 352 — 360). 

5.36. Remarks on Selz' Theories. I have devoted some space to a 
summary of these experiments as I consider them of interest for the 
matter in hand (see 5.5), and I believe that they are comparatively little 
known among philologists. This applies especially to the "gradual for- 
mulation". The two other types are more familiar, owing to their defi- 


nition by Paul and Wundt. Whether the analysis and classification 
given by Selz are correct or not, one thing appears to be established: — 
sentences are formed in more ways than one. I shall return to this 
point later on. 

With regard to the frequency of the various types in ordinary dis- 
course it is impossible to say anything with certainty, but it would 
appear likely that the gradual type is rather more common in everyday 
speech than in experiments of the kind made by Selz. With one reser- 
vation, however: Selz shows us the processes retarded by difficulties; 
the observers have problems to solve which occupy their attention with 
regard not only to import but also to verbal formulation. In ordinary 
discourse, we only occasionally have to pay any attention to the verb- 
alization, which is left to lower centres. The verbalization is practically 
automatic, and the syntactical structure is correspondingly simple. 
Anal5i;ic or gradual formulation would appear to be most natural in 
cases where the referent is familiar enough to be immediately grasped as 
a total; while the synthetic and the gradual types would appear to be 
specially suited for the solving of problems. The referent then requires 
to be thought out. In writing, Selz thinks the synthetic type more 
common, which sounds probable. — On other similar results, see Selz 
II 362 sqq., C. Biihler 183. 

5.37. Linguistic Automatism. The complex series of mental events 
involved in sentence formulation is made the object of conscious atten- 
tion only when, for some reason, the processes are retarded. The 
speaker may then proceed like a careful writer who selects his phrases 
with regard to their grammatical correctness, their appropriateness to 
the topic, and the intended effect on the listeners. 

In normal easy speech this is not so. Most links in the chain are 
automatized through long habituation, and the initial impulse to speak 
normally releases the speech process. We are not conscious of the 
movement until we have already spoken, nor are we conscious of con- 
structing the phrase, although it issues in perfect syntactical form. We 
sometimes even learn our thought from our words, when the impulse 
preceding speech is so vague that it is differentiated and recognized only 
through its effect: — the sentence which is pronounced. We go direct from 
thought to words, or even from the event which provokes an utter- 
ance to the utterance itself. We learn that we are thinking by hearing 
ourselves speak (Delacroix 404 — 408). 

112 GUSTAF STERN 5.37 

Even when we experience a precise impression and an impulse to 
express it, the words used are not expressly willed, they arise automat- 
ically. Speech, like other complex actions, is based on a substructure 
of reflex actions accumulated gradually from childhood. The conscious 
mind only releases the process. The performance of the first phase 
releases the second, and so on. In most cases we only perceive 
the point of departure, the signal that determines the act in time, 
the correlations which determine it in space, and the objects and 
persons to which it refers (Delacroix 361 — 362. Similarly Muller-Freien- 
fels, Vorstellen 415 sqq.. Head, Brain 46, 426 — 428, and Aphasia I 440 

Automatism and conscious control are equally indispensable. With- 
out purpose and intention, the mechanism of speech is nothing but 
automatism and repetition. Without automatism, the finality remains 
nebulous. To speak and to comprehend speech is to use the most 
profound processes of thought as well as an extremely famihar mech- 
anism of intellectual and sensori-motor habits (Delacroix 370, 361). 

In speaking and in comprehending speech the process constantly 
fluctuates between more and less conscious control and interference 
with the automatic mechanism. Easy phrases in our own language 
are produced and understood with a minimum of conscious effort. 
When the mental content is more difficult to formulate in words, the 
speaker has to turn more of his attention to the verbalization and the 
syntax, and similarly, in comprehending speech, the listener has to 
reflect on the meaning of words and attend consciously to their inter- 

Delacroix (362 sqq.) accepts Head's results as indicating the element- 
al functions out of which the perfect speech function is composed, and 
consequently assumes the existence of four groups of automatic functions. 

The first of these is the formation of auditory and articulator y hab- 
its, and the construction of the corresponding images. The establish- 
ment of a co-ordination between audition and phonation is perhaps 
prior in time. This is the first serious task of the child in learning 
to speak, and its difficulty is apparent from the fact that it takes 
about two years to learn. If this function is disturbed, we have the 
form of aphasia termed by Head verbal defects. Superior functions 
of language may still be in order; the rules of the game persist, but 
the counters are missing or damaged. 


Secondly, words have meanings, to which they are linked. The hab- 
itual connection of word and meaning constitutes a second group of 
automatic functions. This connection is destroyed in nominal aphasia. 
The counters are there, in their articulatory or visual form, but their 
value in the game is lost. 

Thirdly, the formation of relations has become an automatic func- 
tion in normal speech, based on habit and memory. When this func- 
tion is disturbed, a person may be able to use words, but not to com- 
bine them. Pick calls this agrammatism; it results in a telegraphic 
style, purler negre (Head's syntactical defects). 

Finally, the speech function involves the ability to grasp the ulti- 
mate meaning of phrases apart from the verbal meaning, to understand 
the final aim and purpose of action. This function may also be more 
or less automatized. If it is destroyed, language loses its significance. 
If verbal automatism persists simultaneously, speech will exceed 
thought, and the patient will speak without saying an5rthing: there are 
more words than ideas. 

All these automatisms are so many economies of effort. They have 
once been conscious, but as they become habitual they come to be 
performed without conscious effort and permit us to grasp instantly 
in the form of a swiftly envisaged pattern masses of detail which are 
thus handled as units. If so many elements of the speech-function 
were not automatic, we should constantly be confronted with the task 
of consciously performing an overwhelming number of acts. 

The remarks on Head's theories above (5.27) and the account of 
Selz' experiments will have shown that we can make some additions 
to Delacroix' list. The first is the abihty to grasp relations in time 
and space, if van Woerkom is right on that point (5.27). The second, 
more important and more clearly demonstrated, is the existence of 
automatic sentence schemes. 

The importance of sentence schemes will have become clear in the 
preceding paragraphs. With regard to their formation, Selz states 
that they are the result of successful solutions of a task for which 
the use of language was the necessary instrument. When they have 
become sufficiently familiar, they are actualized automatically in the 
presence of similar tasks, simply through the awareness of the task 
and the purpose to be attained: the combining (Zuordnung) of pur- 
pose and instrument (= the sentence scheme) is due to a prior use of 

Goteb. Hogsk. Arsskr. XXXVIII: i 

114 GUSTAF STERN 5.37 

this instrument for the same purpose. It is what Selz calls routine- 
mdssig determinierte Mittelaktualisierung. 

In other cases the process is not entirely automatic. The definite 
apprehension of the purpose [die konkrete Zielsetzung) leads to a find- 
ing of the instrument either through inference, or on the basis of 
previous experience. This is determinierte Mittelahstraktion mit nach- 
folgender Mittelanwendung. 

Finally, cases occur in which the apprehension of the purpose grows 
out of the discovery that such a purpose can be attained by means 
of some process more or less accidentally found. Or, as Selz formu- 
lates it, "die Zuordnung von Ziel und Mittel beruht auf einer der kon- 
kreten Zielsetzung vorausgegangenen Abstraktion eines Wertwirkungs- 
zusammenhanges, namlich des Verhaltnisses zwischen einem tatsach- 
lichen Vorgang und einem durch ihn bewirkten unbeabsichtigten wert- 
vollen Erfolg". This type is not, I think, so important for the matter 
in hand (Selz II 680 sqq.). 

On this theory, it is the conscious apprehension of the purpose that 
releases the automatic functions adapted for the accomplishment of 
the purpose (Cf. S. Meyer, Zs. f. Psych. 65, 97: bei der Mechanisierung 
wird die Zielvorstellung zur einheitlichen Vertretung beliebig grosser 
Reihen). If the means had to be laboriously found and attended to 
detail by detail, we should not be able to perform more than a frac- 
tion of what we really do every day. The helplessness of aphatics in 
many situations of daily life shows our dependence on the undisturbed 
possession of these acquired automatisms. 

5.4. Temporal Relations in Speech Production. The two points of 
interest for the present study are the following: (i) What can we say 
concerning the temporal relations of the psychic acts in the process 
of speech production? (2) At what point in the process do words 
arise in the speaker's mind? 

Van Woerkom (J. de Psych. 724) states that the development of an 
idea for the purpose of verbal expression passes through the following 
phases: (i) the conception of the total (compound) idea; (2) a psychic 
process of analysis and synthesis in time and space; (3) the conception 
of the sentence scheme, without verbal symbols; and (4) the choice 
of words. Pick inclines towards a similar view, although he expressly 
states that the various processes may follow in any order (Pick, Sprach- 
stor. 232 — 234, 247). Delacroix thinks that Pick and van Woerkom 


are too fond of an hierarchical order, but he, too, assumes that the 
initial phase is always "une aperception simultanee et d'abord syn- 
thetique" which is subjected to a "decomposition successive" (Delacroix 


It seems to me that in view of the results gained by Selz such an 
opinion can no longer be upheld. Whatever we may think of his clas- 
sification, the reports of his observers show a variation that is too 
well established for any one form of mental formulation to be the only 
norm. A pre-verbal phase of "gedankliche Gliederung" follows on the 
preliminary adjustment (see below) only in analytic formulation, while 
in synthetic formulation the words appear in support of the succes- 
sively arising thoughts and the advancing mental differentiation. In 
gradual formulation the sentence scheme collaborates from the begin- 
ning, providing a frame-work for the mental differentiation. 

Head definitely opposes any theory of a fixed temporal succession 
of the elemental functions composing the perfect speech function. 
"Speech, examined introspectively, appears to be a progressive act, 
which may be analysed into events appearing at separate moments of 
time. As a gun is aimed, the trigger pulled and the cartridge ex- 
plodes, so it would seem as if we first think of what we want to say, 
then select the terms in which to express it and finally embody them 
in words and phrases. But this is certainly a misleading and fallacious 
method of stating what actually occurs. An act of speech comes into 
being and dies away again as an alteration in the balance of psycho- 
physical processes; a state, never strictly definable, merges into another 
inseparable from it in time. When this transition is interrupted and 
the evolution of a perfect response is prevented by physical causes, 
fresh integration becomes necessary and new phenomena appear. These 
in no way represent temporal elements in a series of normal events. 
Unimpeded symboHc formulation and expression cannot be analysed 
into a sequence of semantic, nominal, syntactical and verbal processes 
which normally follow one another in time . . . Had these reactions 
corresponded to the constituent parts of an orderly sequence in nor- 
mal speech, disturbance at some definite point in time would have 
prevented the development of all those processes which followed later 
in the series. This is certainly not the case; these disorders of speech 
do not reveal the normal order of psychical events" (Head I 509). 
At a later page Head recurs to the question, quoting Pick and van 


Woerkom, and states that such detailed analysis is a mistake "because 
an act of speech does not come into being and run its course in this 
diagrammatic manner" (1. c. 530). 

Head's view is, I think, corroborated by what Selz has found for 
the normal processes. The assumption of a strict succession: com- 
pound total idea + mental differentiation -f linguistic formulation + 
verbalization, cannot be considered as tenable. The various element- 
ary functions are operative more or less during the whole period of 
preparing for expression. Cf. Spearman's analysis of growth on dif- 
ferent levels in the process of comprehension, which, I think, is also 
appUcable to the production of speech (cf. 6.4, and Selz II 318 — 319)- 

However, we need not completely give up all attempts at estabUshing 
a succession of events in speech production. I think that we should 
assume that the first phase in the process is always the adjustment to 
the extant situation (the Stellungnahme) to which Pick especially has 
called attention (Pick, Sprachstor. 138 sqq.). 

Pick quotes a number of writers who have emphasized the adjust- 
ment (Stellungnahme) as essential for speech. C. & W. Stern (Kinder- 
sprache, 179 sqq.) define the sentence as "der Ausdruck fiir eine 
einheitUche (vollzogene oder zu voUziehende) Stellungnahme zu einem 
Bedeutungstatbestand", and state that "Vorstellung nur das indifferente 
Vorhandensein eines Bewusstseinsinhaltes gegenstandhcher Art, Stel- 
lungnahme dagegen ein alternatives Verhalten eines einheitlichen Sub- 
jektes, ein Anerkennen oder Leugnen, Zustimmen oder Ablehnen, 
Wiinschen oder Fliehen, lyoben oder Tadeln bedeutet" (Pick 1. c.).^) 
Miiller-Freienfels (Einfluss 392) defines Stellungnahme as all the reac- 
tions arising in our organism in response to an impression from the 
outside, or to its reproduction. Especially do feelings and emotions 
belong to it, together with volitional elements (Cf. Cassirer, Phil. Ill 
317 — 318, quoting Hughlings Jackson, Brain 38 (1915) p. 168; Boll 694). 

I have assumed above (2.13) that speech (perhaps the sentence, at 
any rate the complete utterance) possesses a fourth function, not found 
in the meanings of isolated words, viz. the purposive function. In agree- 
ment with this assumption we now find that speech also expresses 

^) Compare Gardiner's sentence definition: "A sentence is an articulate sound- 
symbol in its aspect of embodying some volitional attitude of the speaker towards 
the listener". And his definition of word: "A word is an articulate sound-symbol 
in its aspect of denoting something which is spoken about". Gardiner 355. 


more than the isolated word: it expresses theadjustment of the subject 
to the situation in hand. Indeed, this would appear to be an indispens- 
able condition if speech is to perforin its purposive function. 

We may regard a situation calling for speech as setting to the indi- 
vidual a task which he has to perform with the help of speech. I have 
defined speech as an instrument for the promotion of purposes. The 
employment of an instrument for any purpose naturally pre-supposes 
that the user of the instrument adjusts himself, and as far as possible 
the instrument also, to the demands of the task and the intended pur- 
pose. In whatever way the verbalization is prepared, it seems to me 
that an adjustment must precede it.^) 

In many cases it is evident that the choice of the sentence-scheme 
must be preceded by an adjustment to the situation, that is, especially 
to the referent and the interlocutor. If we were not adjusted, before 
we begin to speak, for instance, to making a concession, we should not 
get the correct intonation from the beginning of a concessive clause. 
(Pick, Sprachstor. 240). Similarly with regard to speed and pauses, 
unusual word-order employed for reasons of emphasis, the use of meta- 
phorical or otherwise not habitual expressions, and so on. We are 
differently adjusted to different languages that we may happen to 
speak, to different keys in performing music, etc. (1. c. 236 sq.).^) 

The process of speech is thus always controlled by a subjective at- 
titude on the part of the speaker. It is not necessarily renewed for 
each sentence, but may remain essentially the same throughout a 
long speech or conversation. 

My conclusion is that the production of speech normally begins with 

^) For Head, this phase falls outside the scope of his investigation. For Selz, 
with the more or less artificial setting inseparable from experimental research, 
there could be no volitional attitude towards the task; the attitude could only 
be that of complying with the instruction. From the biological point of view, 
whether phylogenetic or ontogenetic, the adjustment is a very primitive func- 
tion, originating in the physical reaction of the animal to changes in the environ- 
ment, pleasant or the reverse (Cf. Rignano i sqq.). According to Pick, the Stel- 
lungnahme corresponds to Marbe's and Messer's Bewusstseinslage , or conscious 
attitude (cf. Clarke, Am. J. of Ps. 22). 

^) Cf. Pick, Sprachstor. 229 sqq., Selz II 367, and Froschels, who points out 
that a practised speaker inhales a suitable amount of air in his lungs before be- 
ginning a sentence. He has thus some notion of what he is going to say. A be- 
ginner makes mistakes in this respect, as do some aphatics (Froschels 91). 


an adjustment, comprising determining tendencies arising from the 
intentional awareness (Zielbewusstsein) .1) 

The final phase of the process is normally the completed verbaliza- 
tion of the speech. With regard to what comes between these two 
phases, no fixed temporal relations can be ascertained. The psychic 
processes interlace in a way that is probably influenced by the tem- 
porary mental state of the speaker and by the nature and difficulty 
of the task; and since these factors vary from one instance to another, 
their effect likewise varies. 

With regard to the second question propounded above, the words 
may, according to Selz, arise at any phase of the process (similarh- 
Valkenburg 20), and the same is true of the sentence schemes. It is not 
even necessary that the schemes should be prior to the words, although 
they often are, and even prior to the concrete mental content that is 
to fill them (as in gradual formulation). 

The appearance of words at an early stage seems explicable. No 
doubt the attention of the speaker is directed mainly, as has been 
stated before, to the referents he is discussing and to the ultimate 
purpose of his speech. But for this purpose speech is the necessary 
instrument, and the rapid and effective working of the instrument, 
that is to say, the rapid finding of suitable words, is a preHminary 
task that cannot be neglected. Also, our notions are so intimately 
bound up with words (5.53) as to have become practically inseparable 
from them, so that very likely, even if not necessarily, any emerging 
thought will, if sufficiently definite in character, call up a word as an 
almost simultaneous concomitant. 

The most that can be said is that the actual utterance of words in 
sentence form comes as the last phase of the process. In easy, normal 
speech, the sentence seems to spring from the impulse to say something, 
that is to say, from the adjustment to the situation (cf. Pillsbur^- 
Meader 195). The intermediate processes are telescoped into such 
an infinitesimal space of time that they are unnoticeable. No doubt 
this is to some degree due to the fact that the sentence schemes em- 
ployed in everyday Ufe are comparatively simple and few in number, 
and are habitual enough to be without difficulty apprehended as units. 

^) Cf. Ach's definition: Determinierende Tendenzen sind Wirkungen . . . welche 
eine Beeinflussung des geistigen Geschehens im Sinne der Zielvorstellung nach 
sich Ziehen (Ach 120). 


There is, of course, no proof of the assumption that the processes ob- 
servable in artificially retarded instances remain unchanged in the 
normal process. They may be merged in more comprehensive acts. 

With regard to normal speech, we cannot indicate any definite phase 
of speech production at which the words arise in mind, and we are 
thus unable to say anything about the precise psychic situation in 
which sense-changes may originate. 

The results of this investigation into the temporal relations of the 
processes in speech production may appear somewhat meagre, but they 
are not without interest. 

A sense-change may be said to occur at the moment when the word 
arises in mind and is connected, as its symbol, with a sufficiently deter- 
minate item of mental content, with which it has not previously been 
connected. Since this may happen at any phase of development, an 
analysis of sense-change has to reckon with the whole complex of 
normal speech functions as being, potentially, operative at the moment 
of change; and, moreover, not only with the context really expressed 
in the surrounding words, but with any other perceptual or mental 
context that may happen to exist simultaneously. The polysemy of 
words connects most of them with many different items of experience, 
and many different spheres of experience, any of which may act as 
stimulus for their evocation. All these circumstances, taken together, 
provide an infinite variation of contexts in which the word may appear, 
subjected to a variety of influences, by which its meaning is moulded 
to suit the momentary purposes and environment. 

The variations of mental content that we wish to express are infinite, 
and the limited — even if very large — number of words and habitual 
meanings is not sufficient to meet the demands we are constantly 
making for adequate expression. In such situations the extreme 
variability of the mental processes preceding speech, and the infinite 
possibility of combinations of words and meanings, provide us with 
the means of adapting speech to our needs and purposes in each single 
case: we are able to make the words express mental content that they 
have never expressed before. 

The exploitation of these possibilities of expression is, naturally, 
whoUy dependent on the corresponding adaptabiUty of the process of 
comprehension: it would be useless if the hearer were not able to follow 
the speaker's meaning, even when it departs from estabUshed habits 


(see ch. 6, especially 6.33 and 6.36). The lack of fixed temporal rela- 
tions, and the consequent flexibility of the speech process, is thus of 
the greatest service to us, both as speakers and as hearers. 

5.5. The Relation of Speech to Thought, and the Selection of Words 

in Speaking. 

5.51. The Relation of Speech to Thought. Discreteness. I have 
discussed in the preceding sections the ordering of words in coherent 
speech and the processes involved therein. One point of view has 
been left aside: the interrelation of speech and thought. Different 
languages differ in this respect, but certain general principles can be 

The arguments concerning the definition of meaning have already 
shown that the mental content of any user of language is by no means 
completely expressed in the meanings of the words he may pronounce 
or listen to. There is always a great quantity of context (3.16, 4.16), 
forming perhaps the larger portion of our total mental content. This 
fact has, I think, been lost sight of during the controversies concern- 
ing an assumed paralleUsm of thought and speech. (See for instance 
Funke, Sprachform 66, 99 and Engl. Stud. 57, 164 sqq.). 

"Acts of free associative thinking cannot be expressed completely 
in words; the whole process is inherently illogical, intuitive and punc- 
tuated by irrelevancies. As soon as we attempt to express our thoughts 
even to ourselves, we re-arrange them and drastically prune away 
redundant and incoherent features. For the purposes of articulated 
speech or writing this process is carried out still more ruthlessly, and 
we strive to cast our thoughts into a form that is not only comprehens- 
ible to ourselves but to our hearers; the results of unrestricted think- 
ing are refined and ordered in accordance with logical canons." (Head 

I 513). 

Head's statement is somewhat forcibly formulated but is no doubt 
right in principle. Spearman points out that the items of mental 
content named by words attain thereby a higher degree of fixity and 
deter minateness, but this result of verbalization "has to be purchased 
at a heavy price. For it entails, and in its effects is intimately blended 
with, the further character of discreteness. By this is meant that the 
concepts nowhere cover the field of cognition continuously, but only mark 
out certain points in it more or less widely separated from each other. 


A contrast may be drawn in such respect between conception and 
perception, taking as an instance the comparatively simple case of visual 
quality. As regards a percept, on the one hand, this can easily fall 
anywhere throughout the whole continuum from greatest brightness 
to greatest darkness, or from extreme red to extreme violet, or from 
any one size, shape, and duration to any other. Between no two varie- 
ties does there exist any inaccessible interval, however microscopic. 
Quite otherwise, on the other hand, is the range of visual quaUty as 
rendered in concepts. So rendered, it is expressed by single words or 
set phrases that have very large intervals between them. When a 
person ceases to conceive a colour as "blue", he usually jumps straight- 
way to "green", and thence again to "yellow" (Spearman 264 — 265). 

Similarly in ordinary discourse when expressing compound ideas 
by means of words. First, all elements and shades of the original 
thought cannot possibly find expression in speech. Our mind is much 
richer than language (Miiller-Freienfels, Einfluss 397 sqq.). Secondly, 
the collective meaning is far from being exhausted by the sum of word- 
meanings. "The passage across from either to the other requires . . . 
a separate cognitive operation — which can even upon occasion pre- 
sent formidable difficulties. The cognitive field may, then, be compared 
to an ocean studded with icebergs. Over much the larger portion, 
including not only sensation, but most thinking also, it is still fluid. 
Only dotted here and there, has the thought frozen into verbo-con- 
ceptual rigidity" (Spearman 276. Cf. Marty, Unt. 145, Feldkeller 287, 
Pick, Sprachstor. 186 with quots. from other writers, Delacroix 121 
sqq.; Cassirer, Phil. I 20 sqq.). 

We find similar opinions expressed by Laguna, who also devotes a 
passage to the question as to what items of cognition are picked out 
by a language for naming. "Not everything in the world has a name. 
Even in our modern world, which has been combed over and teased 
out by the subtleties of civilized language, there remain aspects and 
nuances which escape fixation in speech; which are not only directly, 
but even indirectly indescribable" (I^aguna 272). Since language is 
an instrument for the promotion of purposes, and for the co-ordina- 
tion of group activities, those features especially of the objective 
world are named "upon which co-operative action -pivots. Whatever 
is peculiar to the individual does not need, nor can it receive, spe- 
cification in language. Moreover, what may be in common to many 

122 GUSTAF STERN 5. 5 1 

individuals merely as individuals, because they are alike, cannot be 
effectively specified by language, except in so far as it excites common 
interest. I may specify verbally features of my world which are equally 
features of yours; but unless such verbal response on my part calls 
out some answering behavior from you, directed to what I have 
been talking about, there is nothing to mark the success or failure of 
my speaking. It is only through the convergence of action upon the 
objects of verbal response that language becomes standardized, and 
its terms freed from ambiguity and vagueness of meaning" (1. c; 
cf. Cassirer, Phil. I 253, 255). 

Different languages differ greatly in this respect: one language may 
discriminate dual and plural, another not, and so on; any comparison 
of two languages will furnish numerous instances (Gomperz II. i. 59, 
note 11; Pick, Schw. Arch. 12, 108 sqq.). Objects which are of espe- 
cial interest to the speaking community tend to receive a plurality of 
names, each indicating one aspect, or function, or phase of develop- 
ment, or use etc. of the referent. (Cf. Hocart, Br. J. of Ps. V; Jaberg, 
Arch. 136, criticizing Tappolet, Arch. 131). 

In discussing the correspondence of speech to thought, we have thus 
first to note that only part of the mental content at any moment is 
expressed by words; what items are selected for naming by a language 
as a whole, is a point that has already been touched upon. In the next 
paragraph I shall discuss the question as to what items of content are 
selected for verbalizing in any individual case of speaking. A third 
detail of the problem is the correspondence between the verbal forms 
actually used in speech and the meanings that they express. 

I have pointed out above (4.27) that it is a mistake to suppose that 
a complex expression always denotes a complex referent, or a simple 
expression, a homogeneous referent. With regard to the relation of 
word to meaning there are also some discrepancies. Thus, at the mo- 
ment of formation, a "passive" noun like committee is no doubt felt as 
designating its referent as "the person to whom something is committed", 
but such a person has also other characteristics; these may, and will 
in many situations, predominate in the speaker's attention, and the 
"passive" element, that is to say the apprehension that the referent 
as one to whom some business is entrusted, may sink into the back- 
ground. The meaning is "adequated" (see ch. 14), as is shown in this 
word by the shift of stress. The composite word then expresses a ho- 


mogeneous meaning. This often happens with derivations and com- 
pounds. It does not of course prevent the meaning from being occa- 
sionally specialized in such a manner that the original relation is again 
apprehended (4.265). 

We have also perhaps to assume that in the case of words hke point, 
head, cause, new, etc. (cf. 4.265) the meaning may be specialized in 
such a way that a relation is apprehended: the referent is the head 
of something, the cause of something, and so on. But does this ele- 
ment belong to the word-meaning, or is it an educed relation, and 
consequently to be reckoned as context? I think the latter interpre- 
tation is more Hkely. 

In the case of flexional forms the endings of which are in current use, 
an adequation like that just described is not possible, except in a few 
instances where an inflected form acquires adverbial force and is sepa- 
rated from its stem word. 

But with the exception of these clearly defined cases I believe that 
we have to assume that each word, and each part of a word that is still 
manipulated as an independent linguistic unit — it will be remembered 
that I am dealing with the meanings of words, not with the meanings 
of phrases — corresponds to some item of mental content. If we did 
not make this assumption I do not see how there could be a science 
of meanings. And the fact that stems, endings and prefixes, all denote 
a more or less constant referent, and consequently express a more or 
less constant meaning, whatever the context, is conclusive proof that 
correspondence between words and meanings is the general rule, i. e., 
that each independent word or part of a word expresses a more or less 
clearly apprehended item of mental content. In normal speech the 
rapidity of the process leaves no time for attending separately to each 
item, and automatism enables us to handle complexes as tmits. 

Further, in spite of the discrepancies already touched upon, I think 
we have to assume that each word generally denotes some detail or 
characteristic of the total (phrase-) referent. Where a single word 
denotes a complex referent, the latter is thereby shown to be appre- 
hended as a imit; and where a phrase (as land without mountains) 
denotes a single referent, the complexity lies in the apprehension, not 
in the referent (4.27). 

5.52. The Selection of Words in Speaking. In order to communi- 
cate a set of facts [Sachverhalt), a complex referent, the speaker has to 

124 GUSTAF STERN 5.52 

pick out those details of the referent that shotdd be specifically denot- 
ed in order to enable him to form by the combination of correspond- 
ing words a comprehensible and coherent sentence; secondly, he has 
to select words that express his subjective apprehension of the referent 
and are appropriate to his intention. 

The mechanism of the former of these operations seems to be very 
little known. It is, obviously, largely guided by the available stock 
of words and sentence schemes, and without the highly automatized 
employment of these tools it would be practically impossible. 
The words provide us with a set of ready-made counters, correspond- 
ing each to some element of reality, as apprehended through some 
element of content, and the sentence schemes help us to arrange the 
words in proper order. Our faculty of observing, apprehending, and 
thinking has developed hand in hand with our knowledge of speech, 
and runs in traditional channels determined by the means of expres- 
sion furnished by our language. The importance of this fact for the 
perfect functioning of the speech apparatus comes out especially when 
we are trying to use a foreign language with which we are imperfectly 
acquainted. We automatically arrange our mental content in the order 
appropriate to our mother tongue, but we are constantly finding that 
this will not do; the other language demands a different formulation 
of thought, and we must laboriously learn how to do it. 

We have to assume, then, that the obtrusiveness of certain details 
in the total referent, and their importance for the speaker's purpose 
(see Selz II 354), collaborates with the linguistic material (words and 
schemes) arising in the speaker's mind, and determines what items of 
the total referent shall be denoted by words. 

With regard to the second problem, the speaker's selection of words 
that express his subjective apprehension of the referent and are appro- 
priate to his intention, we are at least somewhat better informed. 
The psychic mechanism underlying it is, according to Selz (II 378 
sqq.), an instance of "determinierte Komplexerganzung" more precisely 
"eine determinierte Benennungsreproduktion", or, in Spearman's term- 
inology, an eduction of correlates (see note to 4.265). The connec- 
tion between meaning and word functions as a relation (Selz: Bedeut- 
ungsheziehung) , the meaning is the initial fundament, and the finding 
of the word is thus a case of regular correlate eduction. An explana- 
tion in terms of associative psychology does not, says Selz, give a satis- 


factory result, and does not show precisely why the name of the re- 
ferent is educed, and not any other word that may be associated with 
it (1. c. 380). 

The important point for the present study is the process of selecting 
those names of referents that are appropriate to the speaker's inten- 
tion. I have assumed the adjustment to be the initial phase of all 
speech, involving the purpose of the utterance; from the awareness of 
purpose spring the determining tendencies that guide the selection of 
words, consciously or unconsciously, and inhibit the use of unsuitable 

With regard to the factors influencing the selection of words, in 
addition to, or merging with, the influence of the determining tenden- 
cies, there are several details to be noted. (See for the following Dela- 
croix 411 sqq.). There is first the available supply of words and sen- 
tence schemes. Habit may call to our lips in certain situations, or 
indeed in a great variety of situations, one stereotyped formula or 
exclamation: a frequent occurrence. (Such a formula is sometimes 
only a signal for emotion, cf. 5.21). Mostly, however, we handle our 
material more freely. "Nous employons un mot a un moment donne 
pour les memes raisons que nous evoquons une idee; usage frequent, 
usage recent, exigence affective, exigence logique, interet momentane, 
influence de la phrase que nous sommes en train de faire et des mots 
que nous sommes en train d'employer" (Delacroix 413). The emotional 
tone characterizing our adjustment makes us select words carrying a 
similar emotive tone, or at least not opposed to it. The consideration 
of the listener or reader imposes further restraints on a speaker, by 
the necessity of being understood, or the desire to be effective. "Une 
bonne partie de notre phraseologie spontanee s'adoucit et s'arrondit 
au contact de I'interlocuteur. Nous avons differentes famous de par- 
ler selon les miheux. Nous refoulons, en presence d'autrui, une partie 
de notre vocabulaire. I^es necessites de la communication et de 1' ex- 
pression pesent sur tout langage" (1. c). We choose different words 
when speaking to adults and to children, to educated and to uneducat- 
ed persons, and so on. 

A very important point with regard to the completeness of expres- 
sion necessary for an adequate communication is the presence of con- 
text, external or internal (cf. 6.32), supplying information about the 
referents. A full context has the same consequences as previous know- 

126 GUSTAF STERN 5.52 

ledge — or the latter may be regarded as context; it renders a complete 
verbal expression unnecessary; incomplete, allusive phrases are typical of 
the conversation between intimates who are famiUar with each other 
and with each other's affairs and points of view, and of utterances con- 
cerning the situation perceptually present to the interlocutors. A hint 
may then suffice, where otherwise a detailed exposition might have 
been necessary. This is the special field of contingent meanings. 
Words, gestures (symbols and signals), and the perception of the 
situation, perhaps also of the speaker's actions, collaborate to demon- 
strate his meaning to the hearer. The proportion of words may, on 
occasion, sink to a minimum without giving us the right to regard 
the expressions actually used as defective: they duly ftdfil their part 
in the communication and are perfectly adequate to it; consequently 
the expression is functionally complete. It is only when we turn 
the matter round and apply the syntactical yard-stick that a phrase 
without subject or predicate may be termed defective. Much contro- 
versy has been caused by failure to distinguish these two points of 
view (cf. 6.3 and 6.4; Cassirer, Phil. I 44; and ch. 10 below). 

6.63. The Reaction of Words on Thought. Speech is, according to 
the view that I have adopted, an instrument for the promotion of 
purposes, and a speaker employs the instrument in the way best cal- 
culated to serve his ends. But, like other instruments, speech is not 
entirely plastic: it has certain characteristics of its own which react 
on the thoughts it is used to express. These characteristics are due to 
the fact that speech is an actuahzation of a language system, the norma- 
tive function of which has been mentioned above (2.11, 3.25). 

Verbalization has, as I have already pointed out, the effect of making 
the thought more vivid and definite: "our idea approximates to the 
fixity and independence of a percept" (Ward 296). According to Bur- 
kamp, it is questionable "ob eine begriffliche Einheit iiberhaupt ohne 
ein Zeichen zu setzen und zu behalten ist, wenn dies auch vielleicht 
nicht immer ein Wort zu sein braucht" (Burkamp 223. Cf. Humboldt 
§ 8, Wundt II 251 and 511, Messer, Empf. 108, Binet 106 — 108, Pick, 
J. de Ps. 1923, 891; Cassirer, Phil. I 22 sqq, 42 sqq.). Not seldom the 
word is the most substantial part of the mental content, and it is 
through the word that the fugitive operations of thought receive suf- 
ficient impressiveness and power of perseveration to survive the mo- 


ment and to be reproduced (cf. Feldkeller 288, Spearman 264, and 
Wundt's analysis of the meaning complex; also 3.21, 14.12). 

The child, in learning language, is provided with an instrument 
which helps him also to distinguish particular features in the reaUty 
surrounding him. And not only cognitively: "the current vocabulary 
of every generation and community has its set of stock epithets and 
descriptive phrases with which persons and things, acts and incidents, 
come to be branded and their public status established. It is thus in 
large measure that the feelings and sentiments of the community are 
directed and canalized, that prejudices are fostered, and conventions 
maintained. In acquiring the vocabulary of his day, each adolescent 
youth is being fitted with a set of variously colored spectacles, through 
which he is to look at the world about him, and with whose tinge it 
must inevitably be colored. It is peopled with "reds" and "reaction- 
aries", "flappers" and "lounge-lizards", "live-wires" and "morons". 
It is a world in which "pep" and "efficiency" and "personality" are 
desirable and in which "inferiority complexes" are to be dreaded" 
(IvEguna 288 — 289; Stout II 196, 202; Cassirer Phil. Ill 238 sqq.; Stern, 
Psych. Ill; Paulhan, Revue Phil. 104, 23 sqq., 33 sqq.). 

Although, as stated in the previous paragraph, normally only words 
that are compatible with the determining tendencies are evoked in 
any individual act of speaking, it is not always certain that the word 
used is in every way completely adequate. And since words have 
a definite sphere of use, a definite semantic range, and not seldom 
also a definite stylistic and emotional tone (the result of preponderant 
use in connection with certain referents or in certain contexts), asso- 
ciations not intended by the speaker may accompany the words and 
colour his thoughts in an unexpected manner, or even turn them into 
new channels (Spearman 124). The hearer, too, may gain an erroneous 
view of what the speaker wishes to convey (cf. 6.5). This is 
perhaps especially the case where the formulation is more or less 
automatic, where the phrase seems to spring immediately from the 
impulse to say something, and the speaker does not give himself time 
to think it over before speaking. He contents himself with a feeling 
of direction, an adjustment to a certain purpose. The automatic pro- 
cess may then bring up a formvila not intended. And if the formula 
is what the speaker desires, it may be news to him, too; we sometimes 
learn our thought by hearing ourselves express it. 

128 GUSTAF STERN 5.53 

In this connection I may mention "the censoring of language" 
(Pillsbury-Meader 196 sqq.). Occasionally a word that is evoked will 
in its turn call up associates which are unsuitable to the context and the 
speaker's purpose. The speaker may notice it before he utters the word, 
or perhaps only after speaking. "The second possible setting suddenly 
presents itself just after the phrase has been used, and the associations 
that result are seen to be not at all in harmony with the intention that 
is controlling the expression as a whole" (1. c. 197). One may also 
become aware of a formal deficiency in the spoken sentence. Mostly, 
the judgement passed on the words is of a quite general nature; they 
please or displease, they are felt as being suitable or unsuitable to the 
situation or the auditors or the purpose. Totally unsuitable names 
for the referents are usually altogether inhibited as incompatible with 
the determining tendencies (Selz II 600). 


6.1. Visual and Auditory Perception of Words. Before discussing 
the comprehension of speech it will be useful to say something on the 
perception of words, visual and auditory. 

Experimental research has shown that if letters are combined to form 
familiar words we are able to apprehend during an equal space of time 
a much larger number of letters than if they are arranged without or- 
der. This does not mean that we really perceive more letters in the for- 
mer case: we do not recognize the individual letters but the character- 
istic pattern or configuration {Gestalt) of the word as a whole, compris- 
ing the most elementary configurational qualities, as length, angularity, 
etc. This is the first phase; the second begins with the apprehension 
of some characteristic detail. This detail may give rise to mistakes, if 
another configuration, into which the same detail enters, takes the place 
of the correct configuration in the subject's mind. The apprehension 
of additional details limits the number of complexes capable of satis- 
fying the requirements until finally only one remains (Korte 8i; cf. 
Erdmann & Dodge, Zeitler and Wiegand, quoted by Selz I 122 sqq.; 
Selz II 457, 491, Spearman 252, PiUsbury-Meader 133). 

In the case of coherent phrases the analysis and apprehension of the 
words is facilitated by the context (cf. 6.3). The number of complexes 
that fit into the scheme of the whole is much reduced, and in most cases 
it is limited to one possibility long before all the individual letters are 
recognized (Korte 1. c). This explains the passing over of misprints. 
It is well known that careful proof-reading is a much slower process 
than "reading for sense" (Pillsbury-Meader 129 sqq., 138). 

Similarly the auditory perception of words is much facilitated if the 
pattern of the words is familiar. A previous knowledge of the topic, 
famiharity with the speaker's point of view, and inferences from the 
already apprehended portion of an utterance help a listener to educe 
what he does not clearly perceive, to understand more rapidly and cer- 

Goteb. Hogsk. Arsskr. XXXVIII: i 


tainly the words lie is just hearing, and to educe in advance what will 
be said in the next few words. When listening to a speaker at a distance 
or on the telephone, unfamiliar proper names and numbers are much 
more difficult to catch than other words, because we have in the for- 
mer case no assistance from the context (Biihler, Ber. 103, Paul, Prinz. 
78 sqq., Gutzmann 8, Stout II 128).^) 

Neither in hearing nor in reading do we carefully note each ending 
and particle. We note the main points of the sentence, having often 
no need to be guided by the speaker (writer) in reconstructing the rela- 
tions between items; these are easily educed (Selz II 354) . In telegraphic 
style, the words for many relations are intentionally dropped with- 
out impairing the intelligibility of the text. Bagley has shown that in 
ordinary discourse the supplementation is fullest and most correct to- 
wards the end of a sentence, less so in the middle, and least at the be- 
ginning (Biihler, Ber. 103. See also Moore, Mono. 79). 

The psychological reasons for this reliance on eduction and supple- 
mentation where careful reading or listening would provide the data, 
are to be sought in the greater rapidity of the eductive process, and the 
general purposiveness [Zielstrehigkeit) of the speech processes. The lis- 
tener all the time is directed towards the rapid apprehension of the 
referents and of the speaker's intentions, for which the words are a 

^) Selz (I 168 note) calls attention to the circumstance that our comprehension 
of words is undisturbed by individual variations in pronunciation, although these 
may be considerable. "Nicht individuelle akustische Wortbilder, sondern ein 
bestimmtes akustisches Schema, dessen Eigenart auch bei Verschiedenheit der 
individuellen Sprechweise erhalten bleibt, haben wir uns mit den entsprechenden 
Bedeutungserlebnissen verkniipft zu denken. Die Bedeutungserlebnisse sind also 
nicht individuellen Wortbildern, sondern diesen individuellen Wortbildern ge- 
meinsamen Komplexbestimmtheiten reproduktiv zugeordnet" (Cf. Biihler, Ber. 
94 — 95). This view explains why some variations in pronunciation do not 
disturb comprehension, while variations in other directions, may do so. The 
law that reproduction follows not only after repetition of identical items, but 
also after the occurrence of those that are only more or less similar to the original 
experience, is not an adequate explanation, since it does not explain why modi- 
fications in one direction have no effect on comprehension, while modifications 
in another direction may seriously disturb it. It would be interesting to investi- 
gate the why and wherefore of these facts, and attempt to ascertain what modi- 
fications lead to the word being regarded as no longer the same, and what modi- 
fications are felt as lying within the latitude permitted to the word without losing 
its identity. 


means that he strives to pass over as quickly as possible. To notice all 
verbal details would take too much energy and time, which are re- 
quired for the referents (cf. 6.3). 

6.2. The Comprehension of Isolated Words. 

6.21. The Act of Understanding in General. I have quoted above 
(2.21) Ogden and Richards as stating that "a symbol becomes when 
uttered. . . a sign to a hearer of an act of reference. But this act, ex- 
cept where difficulty in understanding occurs, is of little interest in it- 
self, and the symbol is usually taken as a sign of what it stands for, 
namely that to which the reference which it symbolises refers". 

To understand a sign, then, is to make an act of reference, similar 
in relevant respects to that made by the speaker; or, in other words, 
to realize what the sign stands for. 

This definition applies to signals (2.22). For symbols it must be qua- 
lified. Words as symbols have peculiar properties, and represent their 
referents in a peculiar manner. To understand a word it is not enough 
to make the correct reference, it is also necessary to understand the re- 
ferent. It might be argued that if somebody shows us an object that we 
have never seen before, and the nature and utility of which are entirely 
unknown to us, and further tells us that this object is a gazun, then we 
understand the word, since we make the correct reference. But in connec- 
tion with speech the term "understand" is generally not taken in this 
wider sense. To understand language is to understand by means of 
language, and involves understanding the thing symbolized in addition 
to the symbol (Delacroix 441). It is of course an indispensable condi- 
tion for comprehension that the word be conceived as referring to an 
apprehended referent (Fischer 43, 44. Cf. Gomperz, Sinn 11, 47 sqq.). 

We are here interested primarily in the comprehension of speech, 
that is, of actual meanings. However, isolated words are a more easily 
handled material for experiments, and it will be useful to begin with a 
summary of some results obtained with regard to them. On earlier ex- 
periments, see Biihler, Ber. 107 sqq.; also 4.22 above on the nature of 
isolated meaning. 

6.22. Experimental Evidence. A useful and clear account of experi- 
ments on the understanding of isolated words is given by A. Hermann- 
Cziner (Zs. f. Psych. 92). She instructed her observers to read the word 
shown to them, to reflect on it from the point of view of "what is that?'' 

132 GUSTAF STERX 6.22 

and to report their experiences as soon as they had arrived at a result. 
The observers were thus to study not only the first apprehension of 
meaning, but also the subsequent events and their relation to the first 
experience, their value for understanding, and so forth. 

Before understanding proper, there sometimes appeared a feeling of 
familiarity or strangeness, as well as an observation of the form of the 
stimulus word (cf. R. M. Ogden 227 — 228). In normal use the feeling of 
familiarity is merged in the meaning of the word and does not obtrude 
itself, except where understanding fails to follow, that is to say, when 
we recognize the word but do not remember its meaning — an unpleas- 
ant experience in some situations! See on this point also Messer, 
Arch. 8, 85 sqq., Husserl II. i. 73 — 74; Biihler, Bericht 109 about Be- 
kanntheitscharakter; Selz (II 455) explains the feeling of familiarity as 
"Erlebnis einer Gleichheitsbeziehung zwischen den wahrgenommenen 
Komplexen und bestimmten gelaufigen allgemeinen Komplexcharak- 
teren (von Worten, Buchstaben oder Buchstabengruppen)"; Fischer 
(43, 42) states that the Bekanntheitsqualitdt comes before understand- 
ing and is not a consequence of it, and that if it does not appear the 
word is regarded as Sinn- und Bedeutungslos; Freudenthal 27, Moore, 
Mono. 1 01. 

With regard to understanding proper, three distinct variants appeared 
in Hermann-Cziner's reports. 

/. "Pure" tinder standing. In this case the words are understood as 
they are read, without any definable experience whatever.^) This 
awareness of meaning is described as an undifferentiated unit; the 
observer is not conscious of having apprehended the meaning; it is 
present to him in the word, and he is not conscious of a duality between 
word and meaning. In the experience the word does not refer to the 
referent, but it is the referent, it is not apprehended as a name but as 
the referent itself. 

//. Mediated Understanding {das reprdsentierte Verstehen). In this 
type the meaning is represented by some mental event. The re- 

^) Wasser: Ich verstand es ganz leer, ohne dass ich dabei etvvas gedacht hatte. 
— Wahrheit: Nach dem Lesen war wieder ein Zustand des Verstehens vorhanden, 
welcher eigentlich die Einheit von drei Dingen hx sich enthalt: des Wortes, des 
Ich und der Bedentung; es besteht gar keine Trennung. — P flame: Im ersten 
Zustande war gar nichts davon zu fiihlen, dass sich dies auf mehrere Dinge be- 
zieht: es erschien, als ob ihm eben nur ein Ding entsprechen wiirde (1. c. 92). 


presentation may be imageless, and either a feeling^) or a motor innerv- 
ation.2) In other cases the meaning was represented by imagery, in 
which visual images predominated. Concrete nouns were represented 
by a picture more or less complete, but nearly always general. The 
generality is a generality of intention: only in exceptional cases is the 
image referred to individual objects.^) 

The visual representation of abstract nouns and other grammatical 
categories is either an illustration of an instance*) or an image of some 
object that stands in exterior or intrinsic relation to the referent.'^) 
Fairly frequent are images which do not refer to existing objects, but 
present a symbolic representation of the referent^) (Cf. 4.14). 

A successive presentation of images representing objects falling within 
the range of the word occurred only twice in all the experiments, al- 
though it ought to be very common according to the associationist 

Mediated understanding, like pure understanding, is one undiffer- 
entiated experience, in which referent and its representation fuse. 
The representation does not exist beside the apprehended and compre- 
hended referent, but is from the beginning "objectified". The image 

^) Prinzip: Ich fiihlte etwas Starkes, Unentwegtes, und wusste nichts weiter 
herauszubringen. — Sogar: Hier ist nur etwas Gefiihlsmassiges gegeben: ein 
Ausruf, ein selbstandiges Gefiihl . . . Es bedeutete ein Crescendo (1. c. 93). 

2) Prinzip: Das erste illustrative, sich nicht in Worten abspielende Erlebnis 
war die Innervation einer Handbewegung. Es war etwas Hartes, Festes in der 
Stimmimg des Wortes und in der damit einhergehenden Innervation. — Indi- 
vidualitat: Da verspiirte ich zuerst etwas motorisches. Ich hatte am liebsten 
eine abschliessende Bewegung gemacht, fiihlte sozusagen den Anhub dazu. Das 
geschieht gleichzeitig mit dem inneren Verstehen des Wortes (1. c. 94). 

') Lampe: Das erste war iiber einem Tisch das Bild einer Hangelampe, welche 
ein dunkles Zimmer beleuchtet. Ein sehr bestimmtes Bild, obzwar es nicht lo- 
kalisiert war. — Wagen: Hier tauchte auch sogleich ein Bild auf, ein verschwom- 
menes Gemisch von Wagen und Karren, kein deutliches Bild (1. c. 95). 

*) V erwegenheit : Es erschien ein Bild: jemand springt von einer Briicke lier- 
unter (1. c. 95). 

^) Instinkt: Erst dachte ich an ein Tier, sah anschaulich sein Riickenmark, 
als Ort des Instinktes (1. c). 

*) Gedanke: Als ob ich innwendig etwas sehen wiirde, etwas teilweise Visuelles: 
ein Sich-offnen, ein Sich-entfalten . . . Sehnsucht: Erst komnit ein Bild: die Aus- 
weitung von etwas, als ob sich etwas von einem Punkte zum anderen ziehen wiirde, 
eine Leere ausfiillend. Jeder "Sehnsucht" entspricht ein so gedehntes, zusam- 
menhangend lineares Etwas (1. c. Cf. Fischer, Arch. 43, 52). 

134 GUSTAF STERN 6.22 

is the referent itself. This is true also of the emotional representation. 
The importance of the representative content for the comprehension 
of meaning is not contingent on the adequacy of the imaginal ele- 
ments. This is shown by the fact that the representation is not more 
common or more important for concrete words, and that the emotional 
and symbolic elements are very important. On the whole, Hermann - 
Cziner thinks that the representative images may best be characterized 
as symbolic. 

///. Explicative Understanding. In this type the meaning of the 
stimulus word is apprehended through other meanings, or through its 
relation to other meanings. The meaning of the stimulus word may be 
analysed (a characteristic of the referent, or some individual instance 
of it, may be evoked), or related to other meanings (identified with other 
meanings, apprehended as belonging to a superior notion, or to some 
objective sphere, compared with subordinated, or co-ordinated, or 
otherwise related meanings), or it may be defined. 

This determination through other meanings may fuse with the un- 
derstanding of the word, so that its meaning is apprehended from the 
beginning as thus determined or as thus related to certain relevant 
meanings. This type was comparatively rare. Hermann-Cziner em- 
phasizes that it is a unitary experience, just as much as "pure" under- 
standing. Its complexity is apparent only, and conditioned by the lin- 
guistic formulation: in the actual experience the determinants are not 
added to a separately apprehended meaning, but the apprehension of 
the meaning takes place in and through the determination. 

In other cases, more numerous, the explicative determinants connect 
with the already comprehended meaning, either repeating the appre- 
hension of the meaning in a new form, or adding relations and charac- 
teristics in the shape of awarenesses and constatations. This awareness 
or constatation seems to be an item of what I have called context (cf. 
Husserl's Bedeutungserfiillende Akte, II. i. 38). 

It is further evident from the reports that "pure" understanding 
and mediated understanding are frequently followed by explicative 
understanding, and in these cases the latter is apprehended as some- 
thing additional. The first phase of understanding is apprehended as 
something concluded, to which accrues an awareness of relations, etc., 
so that for the unitary apprehension of meaning is substituted a phase 
of reflexion concerning the referent. The addition is conditioned by 


the observer's feeling that by contenting himself with the first phase 
he has not completely obeyed the instruction. The observers agree in 
ascribing practically no importance for the process of understanding, 
to this additional content.^) This is partly due to the fact that we 
have to do with isolated words, which are not vitally connected with 
any interesting actuality or sphere of knowledge; they have no inter- 
pretative value with regard to a speaker. 

The explicative processes are found to occur especially when the sti- 
mulus word stands in an apprehended relation to the observer's sphere 
of occupation or interest, or when the word has been used in a sub- 
sidiary test series in connection with formulation of judgements, etc.; in 
short, when the word is brought into some context. The comprehension 
is then not directed to the isolated meaning of the word, but the latter 
is apprehended in the aspect determined by the context. It is therefore 
probable that the explicative type is more common in ordinary dis- 
course than in these experiments.^) 

Hermann-Cziner thinks that her results show the insuffiency of the 
associationist theories, according to which comprehension is based on the 
reproduction or on reproductive tendencies of complexes of mental con- 
tent, connected with the meaning (this only fits signals, see 6.21). In 
this respect her conclusions agree with the results of research concern- 
ing the production of speech, as shown in the preceding chapter. Word 
and meaning fuse (cf. 3.21) in a peculiar manner, and this circum- 

1) Fass: Blitzschnell fasste ich das Wort auf; — dann sah ich ein Fass und 
fiihlte mich so rund . . . Was dann folgte, war nicht mehr interessant, bewusst 
fiel mir allerlei ein ... — Gewohnheit: . . . Ich denke nach, wie sich die Gevvohn- 
heiten entwickeln, Intelligenz und Gewohnheit fallt mir ein . . . Aber das Wissen, 
das ich beim Erblicken des Wortes gewann, veranderte sich dadurch iiberhaupt 
nicht (1. c. 104). 

^) See on these points Selz I 3 sqq., especially 83 sqq. Selz gave his observers 
tests consisting in the eduction of correlates (for instance, find a co-ordinated 
notion to fishing. Reaction: hunting). He found that in the case of familiar 
notions, the solution appeared in the form of "unmittelbare Wissensaktualisierung"; 
in the case of less familiar notions in the form of "sukzessive Wissensaktuali- 
sierung", that is to say, the solution is found with the help of more familiar no- 
tions which refer to the same topic (1. c. 45, 66). Although there is a difference 
with regard to the Aufgabe, there is thus a certain similarity in the results arrived 
at by Selz and Hermann-Cziner: in both kinds of tests the response may appear 
as if automatically, or it may be mediated by other items of mental content (cf. 
Selz n 285 sqq.). 

136 GUSTAF STERN 6.22 

stance will naturally lead to word and meaning being apprehended in 
one act, as psychically the same entity. 

It further appears that the meaning-experience is not built up from 
imaginal elements. Mediated understanding is not the only type, and 
not the most frequent one, and the importance of images does not 
lie in their content, but in their function as symbols. This, too, agrees 
with what has been previously stated (4.142). 

Moore (Mono. loi sqq.) gives the following summary of the processes 
involved in the understanding of isolated words. 

(i) A feeling of familiarity or strangeness. (2) An awareness of the 
possible applications of the words. (3) An awareness of purpose, arising 
when the observers were instructed to report as soon as they had com- 
prehended the meaning of the word with regard to the functions of its 
referent. (4) Images, mostly visual. 

Fischer (Arch. 43, 63) gives a similar table: (i) Akustische Wahrneh- 
mung des lyautgebildes und Gerichtetsein auf den Wahrnehmungsgegen- 
stand. (2) Bekanntheitsqualitat. (3) Richtungsbewusstsein oder In- 
tentionserlebnis auf Bedeutungssphare oder Wissen. (4) Darauf oder 
gleichzeitig Wissen oder Spharenbewusstsein. Eigentlicher Verstand- 
nisprozess damit abgeschlossen. (5) Vorstellungen konnen als Illustra- 
tionen hinzutreten. 

Moore's second item probably corresponds more or less to the Spharen- 
bewusstsein, an awareness of the sphere or direction in which the referent 
is to be sought. Selz (II 117) states that the first reaction to the stimu- 
lus word is often a very abstract general awareness of the genus to 
which the referent belongs. This characteristic is very general and has 
the highest degree of readiness, and so is the most easily and swiftly 
evoked (1. c. 163, 285; similarly Messer). Fischer especially emphasizes 
this item. We are able to understand by being aware that the referent 
belongs to a certain group, even if we do not explicitly relate it to other 
objects belonging to the same group. "ErfassUng ist Einordnung" 
(Fischer, Arch. 42, 362, and 43, 46). 

Moore's third item, the awareness of purpose, is evidently evoked by 
the task set to the observers, which led them to think of some object or 
objects denoted by the word, and their use. But when such a task is 
set, the word is no longer isolated in the strict sense in which I have taken 
the term. It has been provided with a context, a determining tendency, 
which gives the train of thought a definite direction, and thus, in some 


measure at least, also provides the word with a definite referent, a spe- 
cialized meaning. With regard to completely isolated words, I do not 
think that an awareness of purpose enters into their meaning, except as 
a possible item among other indefinite items that may present them- 
selves to the mind in speculating upon the possible meanings of the word 
(if sufficient time is given to observers for such speculation). Images 
seem to be common with isolated words, provided time is given for them 
to develop. Possibly the absence of a definite referent, and thus of 
some point of support for the meaning and the word, leads to an attempt 
to find a substitute in the shape of imagery. 

Other psychologists, as R. H. Wheeler and Tolman, have given their 
observers still more time than Moore did to develop all sorts of con- 
comitant phenomena, which, from my point of view, do not belong to 
meaning. E. Jacobson states that among the concomitant phenomena 
some were felt to carry meaning, others not. "The meaning-associ- 
ates proceed from the instruction given, while the not-meaning-associates 
are external to the instruction; the former indicate the activity of a par- 
ticular determining tendency; the latter indicate the activity of repro- 
ductive tendencies not connected with this determination" (E. Ja- 
cobson 564 sqq., with further comments on this point, and R. M. Og- 
den). In actual speech there is of course never much opportunity for 
pondering on possible associates and correlates, and if they arise they 
are felt to be irrelevant and disturbing. Determining tendencies, on the 
other hand, play a very important part. 

6.23. Conclusions. Moore remarks that "the consciousness of what 
an object is does not necessarily teU us what the object is in terms of use or 
analysis into parts, etc. All detailed information may be suppressed, 
but even then this knowing is both an actual and a potential something. 
It is potential in as much as one may realise that this simple apprehen- 
sion is capable of analysis — that one needs but to split up its apparent 
unity and simplicity and he will become conscious of his entire stock 
of information concerning the object. Nevertheless it is not entirely 
a potential something. Even in its simplest and most unanalysed form 
the consciousness that one has in the presence of an apple would not do 
for an orange or anything else except the apple itself" (Moore, Mono. 77). 

Moore, as well as other experimenters, gave their observers instruc- 
tions which provided a setting or context with which the meaning was 
put into relation. The most common of the mental categories to which 


138 GUSTAF STERK 6.23 

observers assimilated the meaning of the stimulus word were found to 
be (i) for what is it good? (2) of what is it made? Both these categories, 
utihty and composition, are very primitive (Moore, 1. c. Cf. Ach's re- 
mark, Begr. 25, that Zweckdefinitionen are common at the age of six 
or seven; later the genus proximum, an instance, etc. are employed. See 
also Selz II 289, employing similar methods). 

Hermann-Cziner seems to utilize the term understand as indicating 
merely the potential awareness mentioned by Moore, distinguishing any 
further processes as context. Different opinions on the process of un- 
derstanding may be due to different interpretations of the term and 
its extension. 

We may, I think, accept the results of Hermann-Cziner 's experiments 
as in the main correct, since they agree with what we may expect on the 
basis of other data. That the word may seem to he its referent, agrees 
with the observations on the pecuHar fusion of word and meaning re- 
ported above (3.21). It is reasonable to assume that in famiUar words 
significans and significatum should be practically identified. It is 
further reasonable to assume that images may mediate the compre- 
hension of words (cf. above 4.14). 

The third type, explicative understanding, is the one that has been 
especially noticed by earlier investigators. Delacroix states that under- 
standing is a two-fold act. "The elementary representations which we 
arrange by correlating them to each other are already "placed" by their 
relations to other representations which permit us to understand them; 
when I understand that the runner is hurrying towards his goal, I know 
already what a runner is; when I understand that the painter is pre- 
paring his palette before setting to work, I know already what colour is. 
When I understand a phrase I know already the words, the new arrange- 
ment of which gives rise to a new meaning" (Delacroix 438). That which 
"we know already" is clearly the meaning of individual words, and 
Delacroix thus assumes them to be defined by their relations. (Cf. 6.35 
and 6.36). 

Cf. Biihler's distinction between direktes und indirektes Meinen: "1 
mean this", compared with, "I mean that which complies with such 
and such conditions". (Biihler, Arch. 9, 359, also quoted by Fischer, 
Arch. 43, 44). Fischer thinks that the latter is the rtde in the case of 
a first experience, the former in the case of repeated experiences, 
when the relations remain only in the form of dispositions. 


The meaning of an isolated word always remains vague in the sense 
that it lacks a definite objective reference, as well as a setting of actual 
reality. The comprehension remains at the stage of an awareness of 
possible applications of the word, one of which may predominate, or of 
the general concept which the word denotes (4.22). 

6.3. The Comprehension of Speech (Ordinary Discourse). 

6.31. Preliminary. The main difference between isolated words 
and words in speech, is that the former lack vital connection with 
actualities, not expressing the purpose of a speaker, and having no 
definite objective reference (cf. 4.22). In speech, the reference of each 
single word is mostly one among several possible ones, that is to say, the 
words have mostly specialized meaning. We shall have to show how 
it is possible for the hearer to select the meaning really intended by the 

We shall further have to discuss the interrelation of word-meanings 
and phrase-meanings. As will be evident from the following pages, it 
is impossible to separate these two in the analysis. Words do not occur 
in living speech except as elements of sentences, or as one-word senten- 
ces. The constructing of phrase-meanings on the basis of word-meanings 
and other data will have to be analysed. 

6.32. Context: Verbal, Perceptual, and Mental. The Hstener has to 
infer from the speech he is hearing both what referents are intended, 
and what meaning is intended, that is to say, how the referents are to 
be apprehended. Since meaning comprises the objective reference, it 
is not necessary to distinguish the two phases of the process. We there- 
fore ask only, what guides has the hearer to assist him in interpreting 
the meaning of the words as it is intended? 

The answer may be given in one word: context. Context includes the 
immediate verbal context together with the non-verbal means of ex- 
pression spoken of above (2.22 and 4.152), external (perceptual) con- 
text (the situation), and what mental context the hearer is able to 
supply from his own knowledge, experience and observation. The ele- 
ments belonging to context may refer either to the topic, or to the speak- 
er and his attitude to the topic and to the hearer. 

We have to adduce first the general psychological law that if a stimu- 
lus is connected with several reproductive tendencies, that tendency is 
generally realized which is previously more or less actual in mind. A 

140 GUSTAF STERN 6.32 

preceding "preparation" gives to this tendency an enormous prepond- 
erance, while the other, unprepared, tendencies do not attract attention 
(they are "detracted", Poppelreuter 307 — 310). This is especially the 
case when several tendencies are convergent, that is to say, directed 
towards the same end (1. c. 312. Cf. on the whole problem also Selz I 
222 sqq.) 

Words may be regarded as stimuli for the reproduction of meanings. 
Each word is connected with a number of specialized meanings; one of 
these is relevant to the general topic of speech, and therefore in a state 
of potential readiness or preparation. The particular stimulus (the 
word) and the general stimulus (the topic) converge towards that one 
specialized meaning, which is therefore evoked in preference to any 

With regard first to the verbal context (the immediate verbal context, 
as distinguished from the wider verbal or textual context: statements 
in previous sentences, chapters, etc., which I reckon as mental context), 
the words in a sentence provide determining elements for each other. 
The addition of attribute, verb, adverb, and so on, to the subject-noun, 
will greatly limit the number of referents that might possibly be denoted 
by the noun, cutting off, as it were, successively further portions of 
the referential range as not relevant; the noun likewise determines the 
precise meaning of the other words. Onl}' meanings which suit all the 
words in a sentence can be intended by the speaker, since the sentence 
is to express a logically coherent thought. 

In an analytic language the influence exerted by the words on each 
other is largely dependent on the order in which they are arranged. If 
we mix the words in a sentence and read them in an arbitrary order, the 
resulting experience will be quite different (Poppelreuter 328). 

The influence of verbal context has been experimentally demonstrat- 
ed. If the German word Bank is offered to a number of persons, they 
will report varying experiences. But if they are given the pairs Geld — 
Bank, Park — Bank, Gerichtssaal — Bank, the first word restricts the 
associations evoked by the second word to a definite sphere. Only 
those elements that are already "prepared" by the first word tend to 
be reproduced. 

If the second word is unknown, it will be tentatively interpreted as 
belonging to the sphere actualized by the first word. Krankenhans- 
Keratitis: "at first meaningless, then assumed to be a disease; sound- 


ed medical". Afrika-Misitis: the second word was apprehended as 
the name of some place in Africa; other African names were reproduced. 
A new notion is thus assumed to belong to an already actualized sphere. 
If the meaningless word is placed first, it receives its meaning from the 
familiar word following it (cf. Selz I 222 sqq. on similar experiments of 
of varying kinds). 

When an unknown word occurs in a sentence it is comprehended "en 
fonction du schema d'ensemble" (Delacroix 445). The familiar words 
denote each their portion of the total referent, and the syntactical rela- 
tions between them and the unknown word correspond to the actual 
relations between their referents. The referent of the unknown word 
may therefore be inferred through an eduction of correlates. 

The mutual limitations exercised by the word meanings in a phrase 
may be sufficient to make the total meaning perfectly definite, and to 
reduce the number of possible phrase referents to one: that is the case 
with statements concerning unique objects, as the diameter of the earth is n 
miles, the Nelson Column stands in Trafalgar Square. Nouns used in a 
general sense belong to this type: cows are larger than dogs, the rapidity 
of light is greater than that of sound. Nevertheless, I regard such phrases 
as isolated, as long as they are not the expressions of a speaker's appre- 
hension of the referent; in other words, as long as they are not judge- 
ments, but merely propositions (cf. 4.22). 

The non-verbal means of expression (2.22, 4.152) may give the hearer 
additional information concerning the speaker's attitude towards the 
topic or towards the hearer himself. Gestures, or the direction of the 
speaker's glances, may indicate what is being referred to. In print, 
such signs may be replaced by explicit information on relevant points. 
These signs form a transition between verbal context and perceptual 

Perceptual context is often indispensable to the hearer as a comple- 
mentary determinant of meaning (thus in contingent meanings). That 
is the case with most one-word sentences, for instance as used by 
children. For a child "a substantive does not denote simply an ob- 
ject, but all the actions with which it is in relation in the experience 
of the child" (quoted from Bloch, J. de Ps. 18, 710; cf. Jespersen, 
lyanguage 133). "A child's word does not . . . designate an object or 
a property or an act; rather it signifies loosely and vaguely the object 
together with its interesting properties and the acts with which it is 

142 GUSTAF STERN 6.32 

commonly associated in the life of the child. The emphasis may be 
now on one, now on another, of these aspects, according to the exi- 
gencies of the occasion on which it is used. Just because the terms 
of the child's language are in themselves so indefinite, it is left to the 
particular setting and context to determine the specific meaning for 
each occasion. In order to understand what the baby is saying, you 
must see what the baby is doing" (Laguna 90 — 91). Bloch reports 
that two of his children used pa te (par terre) to call attention to an 
object lying on the floor, as well as to announce that the child had 
thrown it there. One child also used pa te as a request to be put on 
the floor himself (L,aguna 89). 

Dependence on perceptual context may be of all degrees. Pick 
(Sprachstor. 155) quotes Bergson (Matiere et Memoire 133) to the ef- 
fect that normally the inferences from context surpass what is under- 
stood from words. Discussion concerning objects around us usually 
requires perception of the objects in order to be correctly interpret- 
ed. There flies a bird is a phrase which we understand, in one way, 
in any circumstances, but it is only when we can refer it to a definite 
set of facts that its precise meaning is understood, and the possibili- 
ties of interpretation reduced to one (cf. 4.25 on contingent meanings).^) 

There is finally the mental context that the hearer himself is able 
to supply, from his previous knowledge of the topic and his experience 
of the speaker and of the topic, whether this knowledge is of long 
standing or is gathered from statements made in preceding sentences, 
chapters, etc. In the case of scientific statements, as in the present 
work, this context is always indispensable. "The most impersonal and 
abstract of disquisitions is written with a whole background of unex- 
pressed "representations", and in a situation of presumptions. If this 
presumed background is not shared by the reader, what is written 
must remain relatively unintelligible to him" (I^aguna 109). 

Similarly in mo§t narrative texts. We may take a sentence like the 
following: "With his rifle on his shoulder he went off towards the for- 
est". If this sentence, occurring in a book, is purely descriptive, it 
has not much meaning over and above that borne by the single words. 
But it is nevertheless not fully comprehensible by itself. We do not 
know who he is, what the forest is, the time and circumstances of going 

^) See in this connection Biihler's interesting exposition of "Steuerung"(Biihler, 
Krise 39 sqq.). 


off, and so on — a number of things which we should have to be in- 
formed of for complete understanding and which we normally gather 
from the wider textual context. But let us assume that he is going 
off into the forest to encounter a dangerous enemy, or is about to start 
on some other risky adventure from which his friends are attempting 
to dissuade him. We have a tense emotional situation, the solution 
of which may very well be given in the phrase quoted. Its import is 
then completely changed. It is no longer merely a case of fitting a 
simple action into a series of events. We have to grasp the effect of 
a fatal decision on an involved situation, a decision perhaps led up 
to by a long series of acts, and in its turn releasing another chain of 
events. Thus a changed context may completely change the import 
of a statement. "The meaning of a word is determined by its total 
setting rather than by the word itself and its immediate setting" (Pills- 
bury-Meader 141; Cassirer, Phil. I 103, quoting Humboldt). 

Elements of the wider context may be supplied to a reader after 
the sentence that is to be interpreted, so that it is understood only 
in the light of what is read on the next page or in a subsequent chap- 
ter (1. c. 141; 6.37). Elements of mental context that are sometimes 
very important for the fulness and vividness of apprehension, are 
those elements of content that I have described in 4.16, the aware- 
ness of sphere or direction, and everything in the way of associations 
that may be called up in mind by the referent and the word, stylistic 
associations connected with the word, and so forth. 

Context is important in figures of speech, as for instance in irony. 
The hearer must infer, from his knowledge of the topic and of the 
speaker's real attitude towards it, that the speaker means the reverse 
of what he is saying; sometimes intonation may be of assistance, but 
that is not alwa^^s the case, and in print it is often not specially indi- 
cated (see further ch. 11). Context is equally important in the case 
of short or shortened expressions: Fire! Quos ego . . . ! 

Even where it is not necessary, we are accustomed to supplement 
from the context instead of noting in detail what we are hearing or 
reading (cf. Spearman 252, 276, 291, Jespersen, Ph. of Gr. 309 — 310; 
and Paul, Prinz. 79 with an analysis of context that agrees with the 
above on all essential points). 

6.33. Active Interpretation. The receiver of speech bases his com- 
prehension on all the relevant elements described in the preceding 

144 GUSTAF STERN 6.33 

paragraphs under the heading of context — the words of the sentence 
themselves, the non-verbal means of expression, the perceptual con- 
text in all its aspects, and his own mental context: previous know- 
ledge of the topic and of the speaker, together with his critical opinion 
on them all. (On the difficulty of this operation, see Thorndike, Read- 
ing; Delacroix 445). 

To begin with, context "prepares" the correct interpretation of the 
words by indicating at least the direction or sphere in which the topic 
is to be sought, and in many cases it indicates the topic much more 
precisely: we know what the other man is talking about. We may know 
something of his attitude to the referent, and we can consequently 
make a guess at the leading ideas in what he is going to say to us. 
We may have a personal independent knowledge of the topic, or an 
opinion on it, which are placed in at least potential readiness and 
contribute to the attitude that we take up towards the speaker's ut- 

The more a hearer knows of the topic, the context, and the speaker, 
the easier it is for him to interpret swiftly and correctly what the 
speaker means, even if the sign should be fragmentary (cf. on incom- 
plete expression of meaning 5.52). "Le degre d'intensite et de pre- 
cision du signe, necessaire a la comprehension, est en raison inverse de 
la comprehension deja presente a I'esprit. Intelligenti pauca! Dans 
un demi-jour nous distinguons et reconnaissons les objets famiUers, 
alors qu'il nous faut le plein jour pour reconnaitre les autres. Un petit 
signe d'une personne dont nous connaissons la mimique habituelle nous 
suffit pour explorer ses sentiments" (Delacroix 448 . Cf. Marty, Unt. 
148, Paul, Prinz. 15, 81, K. O. Erdmann-43, Pick, Sprachstor. 139 
with quot. from Swoboda; Biihler, Ber. 104, 122). 

The actual interpretation of words thus normally takes place in a 
setting that greatly facilitates its rapid and accurate progress. There 
is a constant* interaction between it and all the contextual factors 
enumerated. These furnish, so to speak, a blank form which has to 
be filled in with actual material from the words of the communication, 
and which may be more or less complete (cf. Selz I 200 on Blankett- 
natur der Aufgabe). 

The most important point in the preparation seems to be that the 
hearer should grasp as quickly as possible the general drift of the ut- 
terance, and thus create for his own mental operations a determinant 


that guides their activity in the desired direction, inhibiting non-re- 
levant elements.^) This general notion of the total meaning is some- 
times furnished by the context, and in difficult cases (as in Biihler's 
experiments) it may be delayed until the whole sentence has been 
spoken and the hearer has been able to draw his conclusions from the 

In order to provide the hearer with necessary data, it is of great 
importance that the speaker should as soon as possible indicate the 
general characteristics of the referent, so that it may at least be 
"placed" in a certain sphere (cf. Marty, Unt. 211), even if not definitely 
known. If the beginning of the speech is ambiguous and badly ex- 
pressed, the hearer is at a loss concerning the correct objective reference, 
and may be completely misled (6.37). Selz (I 194) emphasizes the 
importance of an early creation of "ein einheitliches Zielbewusstsein" 
and Stahlin's observers were always striving "sobald als moglich einen 
Gedanken zu finden, dem sie alles Folgende ein- und angliedern kon- 
nen" (Stahlin, Unt. 156). 

The eduction of relations and correlates involved in the process of 
comprehending speech does not wait on the actual pronunciation of 
words, but very often runs on in advance. If, during experiments, a 
sentence is broken off in the middle, the observer will sometimes report 
that he was adjusted (eingestellt) to the continuation. This may be only 
an adjustment, but in other cases it is a more or less tentative or 
definite anticipation of the sequel (cf. Stout II 122 on "connective 
pre-arrangements") . If the expectation is not fulfilled, there is a feel- 
ing of disappointment (cf. 6.37 on mistakes in anticipation). Speech 
reckons with this process of eduction and anticipation to the extent 
of not always expressing what may be inferred. Wegener points out 

1) Cf. Selz II 600: "Aufgabewidrige Assoziationen der dutch die Ausfiihrung 
einzelner Teiloperationen geweckten Wort- oder Sachvorstellungen bezw. der 
entsprechenden Reproduktionsgrundlagen konnen gegeniiber der durcli die feste 
Zuordnung der Operationen bedingten Erregungsleitung in der Regel nicht auf- 
kommen. Eine liickenlose Kette teils kumulativ, teils subsidiar einander zuge- 
ordneter Losungsmethoden bestimmt in streng fixierter Reihenfolge den gesamten 
Verlauf der Verwirklichung einer Zielsetzung bis zur Erreichung des Ziels oder 
zum Aufgeben der Determination und lasst im allgemeinen keinen Raum fiir 
diffuse Reproduktionen". — Maier 343 states that the sentence heard evokes 
"ein logisch-kognitives Interesse"; this is a primary factor and guides the sub- 
sequent process of reproduction in a definite direction. 

Goteb. Hogsk. Arsskr. XXXVIII: i 

146 GUSTAF STERN 6.33 

that when the beginning of an action is indicated, the sequel is some- 
times left undenoted. In the sentence war was declared on the first 
of May, hut there was no battle, the use of hut can be explained only 
in this way, since there is no contrast between the actual statements 
in the two clauses. Another instance is Die Dolomiten sind schon, aher 
schwer zugdnglich. Here, too, the aher can be explained only by a 
situation in which the beauty is considered as a reason for going there, 
but the inaccessibiHty as a reason for not doing so (Biihler, Satz 10). 
The same tendency shows in the circumstance that a hearer often does 
not attend to every ending and form word, but listens to the main 
words of a sentence only, educing the relations between them in less 
time than it would have taken to follow the utterance in detail (cf. 
Feldkeller 289 sqq.). 

Language is an instrument for the promotion of purposes. This de- 
finition must be valid not only for the speaker but also for the hearer. 
He is mostly the object of the speaker's purpose, and he cannot be 
said to have understood a communication if he has not grasped the 
speaker's ultimate intention with regard to himself, the vital connec- 
tion between the meaning of the speech and the actual situation in 
which speaker and hearer find themselves. We have seen that this 
ability is lacking in persons suffering from semantic defects of language 
(5.26). _ 

It will be evident, then, that the act of comprehension is by no 
means a merely receptive function. This is clear from the fact that 
mental content is not directly transmissible, but has to be translated 
into incommensurable symbols, and re-translated by the hearer into 
terms of mental content and reference. In this activity, every re- 
levant factor is brought to bear on the task. The hearer proceeds at 
once to put what he is hearing into relation with what he previously 
knows (^f the topic and the speaker, as well as with other relevant mat- 
ters. He forms judgements, rejects or approves. We have, as Maier 
points out (346) , a large stock of previous experiences by which we can 
test the truth of what we are listening to. 

A completely passive reception of speech occurs, but it is probably 
rare, and we have to assume all intermediate degrees from that ex- 
treme to the other, i. e. a critically hostile attitude that rejects the speak- 
er's statements in favour of a personal view on the topic — as when a 


scholar is reading a treatise on his own pet subject — even if we do 
not include cases where the speaker, to the hearer's knowledge, is lying. 
The hearer may coldly refuse to be enthused by the speaker's efforts. 
We find corresponding phenomena noted by Selz (I 272), when the 
observers detect contradictions between various factors involved in 
the task set to them. Similarly a hearer may find contradictions be- 
tween statements received and elements of the context, which lead to 
his correcting or denying the statement. 

Comprehension of speech, as well as production, requires the parti- 
cipation of the total mental ability. 

The interpretation of each word in an utterance is based not only 
on the word itself, as is the case when it occurs in isolation, but on 
every item of context that can be brought to bear on the task.^) 

6.34. Spearman s Analysis of Comprehension. Spearman (118 sqq.) 
has analysed the construction of phrase- meaning on the basis of the 
individual word-meanings. 

According to Spearman, the phrase meaning is "fed from three 
chief streams. The first of these has its source in the fact that the 
leading characteristics of the initially given individual meanings of 
the words is their state of extreme disintegration. Each of them has, 
in general, been abstracted from many things without which it could 
not exist really. This is particularly important to the extent that it 
occurs with relations. In language, any of these can be presented 
quite apart from one, or even both, of its fundaments; for example, 
such words as "likeness" can be given and understood by themselves. 
Whereas in reality, no likeness can possibly exist save as relating to- 
gether two or more like items. Evidently, then, the understanding 
of the purport {i. e., phrase meaning) of language must include the 
putting of such fragmentary meanings together again. And this re- 
integrative operation in its simplest form is just the process which we 
were discussing in the previous section; the individual meanings are 

1) It is necessary to emphasize this circumstance in criticism of, for instance, 
Marty's and Funke's analysis of comprehension as applied to the theory of innere 
Sprachform, as well as of Wellander's vermittelnde Vorstellungen. These are based 
on the mistaken assumption that a word in actual speech is interpreted in exactly 
the same way as an isolated word (cf. Marty's analysis of Lowe, quoted below 
11.59, and 14. 15). 

148 GUSTAF STERN 6.34 

connected together in such a way as to "constitute" a collective mean- 
ing or purport.^) 

The constitutive relation thus involved has a special, but by no means 
exclusive tendency to be of the attributive class. It is of this class in 
the case of coupling nouns with attributive epithets, as "long-time", 
"violent-blow", "learned-men". Here belong, too, all attributive 
complements, as contained in "the-time-is-long", "the-blow-is-violent", 
etc. To these must be added all attributive adverbs, such as enter 
into "before-now", "extremely-good", or "to-run-quickly". And 
similar are the cases where an epithet or complement, adjective or ad- 
verb, has itself a complex structure. Turning, next, to the classes 
other than attributive, examples are, "dropping-wears-away-a-stone", 
or "know-thyself". The first of these belongs to the causal class, the 
second to that which we have called objective. 

Such constitutive process of understanding, it should be noticed, 
is often only executed to a very incomplete degree. Thus, the purport 
of "round square" can be understood up to the point of presenting 
the first individual meaning as an attribute of the second; but it can- 
not possibly be understood up to the further point of presenting any 
resultant geometrical figure" (Spearman 118 — 119). 

Spearman's second principle is that of supplemented purport. "It 
makes its appearance when some of the words — perhaps even the 
larger portion — have been tacitly assumed by the speaker, and 
therefore must be added by the hearer. Examples are, "Tom plays 
well, but Dick badly" (omitting "plays" after Dick); "the man has 
gone away" (omitting "who was here" after man); "Fire!" (omitting 
"There is yonder a" before "fire") . . . This . . . may upon occasion 
be simply the eduction of a correlate. Usually, however, the operation 
is more complex" (1. c. fig; cf. on perceptual context 6.32). 

Spearman's third source for collective meaning he calls indicated 
purport (see the quotation in 4.265 above). Indication is used "when 
the constituted collective meaning links a relation to only a single 

1) Spearman describes the constitutive relation in the following manner: "Let 
any two items of awareness be symbolized by the letters X and Y ; any relation 
that holds between them, by r; and all three cohering together, by the bracketed 
{X, r, Y). We can then, with linguistic appropriateness, say that X, r, and Y 
are "constituents" of (X, r, Y). In other words, X, r, and Y bear to {X, r, Y) 
a relation which may be called that of constitution" (Spearman 71). 


fundament. For these two individual meanings together give birth 
to their correlate; and in the very act of so doing, they renounce being 
any portion of the purport themselves" (1. c. 121). 

In addition to these three methods of constructing the phrase mean- 
ing, various other operations are involved. "A very large number of 
further eductions usually enter into the understanding of the language 
by way of embroidering the main body of the discourse with all sorts 
of incidental references, commentaries, and criticisms. For example, 
one clause may be noticed to corroborate or to contradict another. 

It thus appears that the understanding of even comparatively brief 
linguistic structures must involve the noetic unit -processes in very 
great number. Whether by the educing of correlates or by that of 
relations, and whether by constitution or by supplementation or by 
indication, every single word has to make its influence felt in mani- 
fold directions. It has to do so, firstly, upon the phrase in which it 
stands immediately inserted; secondly, and through the mediation of 
this phrase, upon the sentence to which the phrase primarily belongs; 
thirdly, through the further mediation of the said sentence, upon more 
complex sentences in which the latter is compounded; thence again, 
upon paragraphs, chapters, and other aggregates of still greater mag- 
nitude" (1. c. 121 — 122). 

It seems evident that the elementary processes described by Spear- 
man play an important part in comprehension. We must assume 
them to be based not only on the words, but on all relevant items of 
context. Whether we should consider it possible to explain fully the 
phrase meaning on the basis of such processes, or if we should assume 
the phrase to have still further possibilities of its own, is a problem to 
which I shall return below (6.38). 

6.35. Buhler's Experiments on the Comprehension of Sentences. 
Biihler's observers were instructed to report their experiences as 
soon as they had completely understood the meaning of the sentence 
that was read to them. The reports say little on the building up of 
the phrase-meaning, concentrating their attention on the moment of 
Understanding, and what happens just then. I shall quote two of 

"Hat Eucken recht wenn er meint: selbst die Schranken der Erkennt- 
nis konnten nicht zum Bewusstsein kommen, wenn der Mensch nicht 
irgendwie tiber sie hinausreichte? " Report: Sofort recht skeptischer 

150 GUSTAF STERN 6.35 

Zustand (wesentlich lokalisiert in meiner Gesichtsmiene) . Der Satz 
war lang, aber gleich iibersehbar (ich konnte den einen oder den andern 
Teil mehr beachten). Es schloss sich eine kurze uberlegung an, die 
ich dem Sinn nach so beschreiben kann: wie E. wohl zu diesem Satz 
gekommen sein mag? (nichts gesprochen). Da kam mir plotzlich mit 
einem Aha! der Gedanke: das ist die bekannte Anschauung, dass Gren- 
zen nur vom tiberragendem aus festgestellt werden konnen. Dieser 
Gedanke war mir gleich ganz klar bewusst, aber ohne die Form, die 
ich ihm jetzt erst gegeben habe (von Worten kam mir nur ein Frag- 
ment wie "iiber"). Darnach fliichtige Erinnerung an einen andern Ge- 
dankengang gegen den ersten gerichtet und an das Wort "prinzipiell" 
gekniipft. Etwa so: prinzipiell lassen sich Grenzen auch feststellen 
etwa von dem Begriff des Erkennens her (von den andern Worten 
nichts gesprochen, aber ich hatte im Bewusstsein ungefahr den Sinn, 
den sie ausdriicken) . Das Nein erfolgte ganz prompt ohne Ablenkung" 
(Arch. 9, 305). 

(Verstehen Sie?) "Man muss so wohl mitleidig als grausam sein, 
um eins von beiden sein zu konnen". Report: Zunachst war ich dem 
Satze gegeniiber vollstandig ratios. Es trat ein Suchen ein, das hier 
den Charakter eines wiederholten Sichvergegenwartigens der beiden 
Teile des Satzes trug; so etwa, wie wenn ich mich gefragt hatte: "wie 
fangt man das an, grausam zu sein, um mitleidig sein zu konnen und 
umgekehrt?" (Das ist nur Umschreibung, in Wirklichkeit nichts ge- 
sprochen). Auf einmal kam mir plotzhch und unerwartet der Ge- 
danke, dass die Ausschliesslichkeit des einen oder anderen Zustandes 
sich selbst aufhebt, dass beide eben nur durch den Kontrast bestehen 
konnen (was durch diese vielen Worte wiedergegeben werden soil ist 
gedanklich ein einziger Akt gewesen). Dann hab' ich mir die Satze 
noch einmal wiederholt, und dabei hatten sie eine andere Beleuchtung, 
ich verstand sie" (Arch. 12, 13). 

6.36. Variations in the Process of Understanding. In drawing con- 
clusions from Biihler's reports we must keep in mind that these sen- 
tences are isolated, in the sense described in 4.22 above. Even when 
making a statement on a unique referent, they are not an expression 
for an individual attitude, and consequently lack a distinguishing 
characteristic of actual speech. 

They are artificial in some respects. The observers were confronted 
with phrases on an entirely unexpected topic. In actual speech, as 


has been pointed out above, we mostly know of what the other party is 
speaking, and have at least a knowledge of the sphere in which he 
is moving. We need not wait for the last words before we begin the 
interpretation; we can start at once, since the "Spharenbewusstsein" 
will guide the selection of meanings correctly. Biihler's sentences are 
all of them expressions for one, very complex, thought. Hence the 
observers were practically compelled to wait until the end of the sen- 
tence before they could grasp with certainty the main topic. 

The reports show that after the perception of the sentence there is 
sometimes a pause. The phrase meaning then often comes suddenly, 
as from nowhere. Biihler calls this "das Aha-Erlebnis" (Ber. 117). 
It is, he says, a "Beziehungserlebnis". A previously known, familiar 
notion is reproduced, and the hearer becomes conscious of a definite 
logical relation between this notion and the one expressed by the sen- 
tence. The relation may be one of identity, similarity, contrast, 
subsumption, causality, etc. The importance of this relation is clear, 
for by the conscious apprehension of the relation, the new notion is 
"placed" in the mental world of the hearer: it is notionally pigeon- 
holed, and is thereby understood (1. c, Arch. 12, 13 — 14). 

Since phrase-meaning is not the sum of the individual word-meanings 
(cf. 6.38) it follows that the phrase-meaning as a totality must have 
been grasped before the "Aha-Erlebnis" can occur. "Das charakteris- 
tische Verstandniserlebnis findet zwischen Ganzem und Ganzem statt. 
Daraus geht aber ohne weiteres hervor, dass das Aufzuf assende zunachst 
ein Ganzes geworden sein muss, bevor es aufgefasst werden kann" 
(Biihler, Arch. 12, 17). It seems evident that this form of understand- 
ing is equivalent to Hermann-Cziner's explicative understanding (6.22), 
but that also it is not the only possible form. 

Biihler also found (1. c. 19) that phrases could be understood without 
any mediating experiences. His observers report, for instance, "Ich 
habe nichts erlebt, als dass ich wusste, was der Satz ausdriicken will". 
Similarly, other experimenters report that the words may appear to 
be the meaning (Spearman 257, E. Jacobson). Cf. the corresponding 
process in speech production, where the words issue immediately on 
the intention to say something. The reproductive effect apparently 
is bound to the phrase as a whole, on a collective apprehension, and it 
reproduces a corresponding total pattern in the hearer's mind (Cf. 
Selz I 98, on das Gesetz der Komplexassoziation, and 1. c. loi with quot. 

152 GUSTAF STERN 6.36 

from G. E. Miiller). In these cases an originally complex process is 
telescoped into one comprehensive act and cannot further be analysed. 

With regard to Hermann-Cziner's third type, representative under- 
standing (6.22) would seem to occur only for less complex mental 
content, but scarcely for whole sentences; at least not sentences of a 
more complex type. 

I think, therefore, that we may establish the existence of two main 
types of understanding, whether the content be simple or complex: di- 
rect understanding and mediated understanding. The former is prob- 
ably typical of simple and familiar meanings, and occurs with complex 
meanings only if they are sufficiently familiar to be grasped as units. 
The latter type occurs in the case of less familiar meanings, as well as 
in complex c h>es. 

Although there are, as far as I know, no investigations of the prob- 
lem, I may perhaps suggest the possibility of further distinctions in 
the process of understanding. According to Selz, we can distinguish 
three forms of sentence formulation, leaving out cases where the sen- 
tence is conceived as an undifferentiated whole (5.32 sqq.). Can we 
make a corresponding distinction with regard to the forms of compre- 

A synthetic type of comprehension would imply that the hearer is 
in possession of all the words of the sentence, and arrives at an under- 
standing of the total meaning by a simultaneous combination of the 
single word meanings. It seems reasonable to assume that this is the 
case with many of Biihler's instances. 

A gradual type of comprehension would imply that the total mean- 
ing is gradually built up as the hearer apprehends the individual word- 
meanings, and is able to add, by their means, one characteristic after 
the other to the total meaning. It is not improbable that a process 
of this type is common in cases where anticipations of what is going 
to be said, play an important part; the actual words gradually deter- 
mine and define this anticipation: they fill in the blank (6.33).^) 

^) Poppelreuter states that if there are, in a mental complex, some elements 
with a stronger and others with a weaker reproducibility, then the former ele- 
ments take precedence, and the reproduction does not necessarily follow the 
strict order of contiguity (Poppelreuter 271). In a gradual form of comprehen- 
sion this would imply that comprehension may crystallize round salient or fa- 
miliar words, and not always follow the word-order. 


In the gradual type of formulation the sentence schemes play an 
important part as guides to verbalization (5.34). Similarly, in com- 
prehension, we expect, for instance, an antecedent pronoun to be 
followed by a relative clause; a not only makes us expect a subsequent 
hut; peculiar intonations indicate question, concession, command, and 
so forth; non-verbal signs are also informative. The anticipation of 
the hearer is probably more or less determined in form by sentence- 
schemes (cf. James I 254, quoted by Pick, Sprachstor. 271). 

An analytic type of understanding would imply that the hearer 
possesses an awareness of the total meaning, and then, with the 
assistance of the words to which he is listening, differentiates its 
elements. This type might possibly occur as an extreme case of the 
preceding: the anticipation would approach completeness, and the 
words would only carry corroboration. 

The discussion above is based on the assumption that the hearer 
aims at complete understanding of what is said. In other cases, per- 
haps more in reading than in listening, the hearer (reader) is satisfied 
with grasping the main points. In very rapid reading one notices a 
section here and there, some prominent single words, and so forth, 
filling in the remainder on the basis of one's own knowledge of the 
topic and of the writer's attitude. The amount of supplementation 
is increased while the content actually apprehended through the words 
is correspondingly lessened (Pillsbury-Meader 143). 

Another distinction may be made between different ways of attend- 
ing, or different purposes in reading. "Anyone may read proof ac- 
curately and know little or nothing of the sense when finished. One 
may read to appreciate certain points in the style of an author and 
get that to the practical exclusion of all else, or, and this is more fre- 
quent, one may read for the content and get nothing or very Uttle of 
the words or have little appreciation of the style of the author . . . 
One may read a book for the author's opinion on some one point and 
get a very complete account of that without learning much of anything 
else. This last becomes partly selective memory, but it is also in part 
a matter of selection and supplementing in reading . . . The selection 
in each of these instances depends in part upon what is selected from 
the stimuli offered, but also in part largely upon the associated ideas 
that are called out or permitted to enter into consciousness" (Pills- 
bury-Meader 144). 

154 GUSTAF STERN 6.37 

6.37. Mistakes in Anticipation. If the latter part of a sentence 
does not continue in the direction that seemed to be indicated by its 
beginning, comprehension will be more or less confused, or there will at 
least have to be a change of orientation on the part of the listener, often 
a very perceptible change of attitude. The following sentence is 
quoted from Stahlin's experiments (Unt. 157): Das allgemein Mensch- 
liche ist die Sehnsucht fiir etwas Grosses angesehen zu werden. One 
observer reported: the sentence at first pleased me, but to my 
horror it went on: to be considered great. Another instance is the 
French couplet: Le monde est plein de fous, et qui n'en veut pas voir 
Doit se tenir tout seul — et casser son miroir — where the last half-Hne 
will cause a sharp break in the line of thought of most persons reading 
it for the first time. (Other instances, see Beckman 15, Marty, Unt. 

Biihler has also some instances of mistaken anticipation regarding the 
continuation of a sentence. After having heard the sentence "Knaben 
pflegen den Kafer an kurzem Bande zu halten, aber an kiirzerem noch 
halten die Fiirsten den Mann", the observer declared: "Ich war am 
Schlusse enttauscht, ich hatte erwartet. . . werden sie von den Frauen 
gehalten". "Diese Erwartungen", Biihler adds, "sind fiir die Erfassung 
des Zusammenhangs vielleicht ebenso wichtig als die Riickbeziehungen" 
(Arch. 12, 83). Such anticipatory constructions occur even when we 
listen or read with great attention (1. c). 

This "Bedeutungswechsel" was noted by Messer (Arch. 90 — 91), and 
it is remarkable that even in his experiments, where single words were 
used as stimuli, the observers state that the change of referent is accom- 
panied by a change in the appearance or "feel" of the stimulus word. 
Thus one observer reports: "Das Reizwort "Mark" erst als Geldstiick 
aufgefasst; dann sofort an Mark Brandenburg gedacht. Auch das Reiz- 
wort, das ich noch fixierte, erschien in anderem lyicht; es hatte einen 
anderen Wirkungsakzent, als der zweite Gedanke eintrat" (1. c). Biih- 
ler quotes similar observations from his experiments (Arch. 12, 21; cf. 
Messer, Empf. u. Denken 102). 

The two different apprehensions may arise simultaneously, or the 
consciousness of ambiguity may even come before the apprehension of 
meaning. One report says: "Ich hatte sofort das Bewusstsein, dass das 
verschieden verstanden werden konne, ohne noch klar zu wissen, wie" 
(1. c. 22, cf. Selz I 244 — 245). 


6.38. Word-meaning and Phrase-meaning. From what has been 
said above on the comprehension of sentences it follows as a corollary 
that the meaning of a sentence is not equal to the sum of the individual 
word-meanings. The semantic interrelation of words and phrases has 
been touched upon in 4.27 from the descriptive point of view, and in 
5.51 from the point of view of the production of speech. To the evidence 
there adduced may now be added evidence from the process of compre- 
hension. Biihler's reports are decisive in proving that the meaning of 
a sentence involves supra-summative elements of the greatest import- 
ance: we are able to understand every word in a sentence, and still not 
understand the sentence as a whole. i) 

Although the matter as yet seems to be insufficiently investigated, 
we should probably regard phrases as configurational structures [Gestal- 
ten; see 2.13 and 4.27). It is characteristic of a configuration that it 
contains supra-summative properties that belong to it as a whole, and 
are lost in analysis. "One finds space-forms which possess more pro- 
perties than their elementary visual sensations; the same is true of melo- 
dies with respect to the tones which compose them, and of intellectual 
processes with respect to the data in which they originate. For it can- 
not be assumed that sensations of color and tone, and meanings of 
single words, are to be considered as 'parts' of space-forms, melodies and 
higher thought processes; since the exact impression of a visual figure 
or of the specific character of a musical motif, and the meaning of an 
intelhgible proposition, contain more than a sum of patches of color. 

1) An amusing instance is given by Otis: Let us define the word, incvation, as 
meaning the increase in the number of feet per second per second by which the 
motion of a body is accelerated. A corollary of this statement is, not that the 
unit of incration is one foot per second per second, but that it is one foot per 
second per second per second. Most educated people will understand each indivi- 
dual word in this statement, but most non-mathematicians will require some 
time and thought for grasping the import of the whole. Otis explains it in the 
following manner. The unit of rate is one foot per second. The acceleration of 
a moving body is the increase in its rate, that is, the increase in the number of 
feet per second which it moves in succeeding seconds. The unit of acceleration 
is one foot per second per second, that is, one foot per second every second. And 
again, the incration of a moving body is the increase in its acceleration, that is, 
the increase (from second to second) in the number of feet per second per second 
by which the motion of the body is accelerated. The unit of incration is the unit 
of acceleration every second, that is, it is one foot per second per second per se- 
cond (Otis, Psych. R.). 

156 GUSTAF STERN 6.38 

tone sensations, and individual word-meanings" (Helson, 36, 347, quot- 
ing Kohler, Die psychischen Gestalten 1920). "It is impossible to 
describe or derive real wholes from a knowledge of parts. Since the whole 
possesses its own specific properties, we can never tell in advance from 
a knowledge of the parts what the whole will be or how it will behave" 
(Helson, 1. c; cf. also Gomperz, Sinn 36 sqq., especially 39 sqq.). 

Helson further discusses the question of the decisive factors in any 
given configuration, with regard to the properties of the whole and to 
modifications within the structure. "Sometimes it seems to be the form 
of the group which is of especial importance in determining the charac- 
ter of the configurational changes; at other times it may be some single 
factor within the group which exerts its influence over the rest; or again 
it may be the internal relations of the parts which determine what 
the whole shall be. And often the configurationists confess that the 
determinants of configurational events are due to a variety of factors 
some of which are unknown" (1. c. 348). With regard to the peculiar 
properties of configurations, some theorists assert that new properties 
belong to the configurations as totalities, and are functions of the con- 
figuration itself. Others maintain "that the configuration is not univoc- 
ally determined whether as to form or as to structure. Thus four dots 
arranged in the form of a square may be seen as two horizontal lines, 
two vertical lines, or two oblique lines at right angles to each other. . . 
the determinant in equivocal cases must be some higher psychical factor 
which unites sensory contents into patterns" (1. c). 

Since phrase-meanings are not among my subjects, I shall content 
myself with these brief indications concerning an interesting problem 
(I refer the reader to the bibliography given by Helson, Psych. Rev. 37). 
It seems that Helson 's remarks on configurations may be applied to 
phrase-meanings. It is impossible to say definitely, for lack of investi- 
gations, whether the properties of phrase-meanings are entirely de- 
rivable from word-meanings in orderly juxta-position, or if we are to 
assume that the phrase has a meaning of its own, accruing to it as a to - 
tality. In the former case, an analysis on the lines indicated by 
Spearman (6.34) would be adequate, and would lead, if properly con- 
ducted, to a complete explanation. In the latter case, such a method 
would have only a partial success, and there would remain elements of 
meaning that would have to be sought for in other ways. (Cf. Wundt, 
Phys. Psych. Ill 555, on das Prinzip der schopferischen Resultanten, 


also quoted by Spearman ii6; Stout II i sqq. on noetic synthesis; 
Delacroix 203: il n'y a pas que des pierres dans une maison. Mais que 
serait la maison sans les pierres? Gomperz, Sinn 39 sqq., esp. 42). 

Wundt and others have emphasized the priority of phrase-meaning 
to word-meaning, and it is indubitably tru^ that the word-meanings are 
"embedded" in the phrase-meaning (Gomperz II. i. 261). But Wundt 
was not right in assuming that formulation is always an analytic pro- 
cess, and Selz has shown that the word may arise in the mind before the 
sentence, even for the speaker. We have to say, then, that word-mean- 
ings and phrase-meaning are mutually dependent on each other. It 
is only as a member of the actual utterance that the word really carries 
this or that (actual) meaning; and, conversely, the phrase-meaning is 
built up on the basis of word-meanings and their interrelations. The 
words symbolize elements of a total referent, which, as a total, is sym- 
bolized by the whole sentence. From the subject's point of view, the 
sentence expresses, through the orderly combination of words, more or 
less completely, his adjustment to the situation and his ultimate inten- 
tion with regard to it. 

From these two points of view, that of the referent and the subject, 
the word-meaning is undoubtedly largely determined by being an ele- 
ment of a definite phrase-meaning. But from the third point of view, 
that of the word, the word-meaning asserts a certain independence. 
The word has a comparatively stable traditional range, and, whatever 
the context, the actual meaning must normally fall within the range. 

With regard to the purposive function of speech, I have assumed it 
to be a configurational quality, and here the word-meanings can only 
act indirectly. This aspect of the matter is very little known (cf. 4.27). 

6.4. Temporal Relations in the Process of Comprehension. It would 
be of the greatest interest to the present study if it should prove possible 
to ascertain the order in which the various processes follow each other in 
the comprehension of speech. We found that in the production of 
speech no definite order prevailed, and it seems as if the variations were 
equally numerous in comprehension. A great many processes are going 
on at the same time. "And particularly remarkable is the fact that 
this simultaneousness appears to happen even in respect of the proces- 
ses which are built up in successive levels. For in such case, the educ- 
tive growth is unlike the building of a wall, where first one layer is 
finished, and then another above it, and so on. It is more like the 


waxing of a tree, which does not first complete its roots, then stem, 
then branches, then leaves, all in succession, but develops all these 
overlappingly. So in the cognitive cellulation also, the lower levels 
are allowed only a limited degree of priority. Whilst they are still 
extremely obscure, the upper levels already begin growing also, and to 
the full extent that their as yet very imperfect understructure becomes 
from moment to moment capable of supporting. This fact has been 
strikingly demonstrated in respect of ideation by the following experi- 
ment. It has been shown that when a person has read through a 
short passage in a book, his subsequent reproduction of it usually 
commences not with any portion as it was read originally, but instead 
with a most compendious awareness of the gist of the whole. The 
explanation given by the investigator himself runs as follows: 'the 
concentrated essence of the whole is the first thing to come up in our 
minds, because it is the one thing that was growing while every other 
thing was being thought' (Henderson, Psych. Mono. 23, 1903). Evidently, 
to grasp the "concentrated essence" implies to understand, not only 
each word as in relation to previous words, but also each clause as in 
relation to previous clauses, and each sentence to previous sentences" 
(Spearman 82). 

With regard to the production of speech, we were able to state at least 
what the initial and final stages normally were: the taking up of an 
attitude and the final verbalization. With regard to comprehension 
we can apparently not do even this (6.36). We may anticipate what a 
person is going to say at the moment when, or even before, he opens his 
mouth, and we may be right. It might be argued that this is not 
understanding speech: it is inference. On the other hand, inference, 
eduction of relations and correlates, and supplementation, are so 
interwoven with the process of understanding connected speech that 
any discrimination is difficult. 

The continuation of the process is, as stated by Spearman, highly 
complex, and the mental content goes on growing at higher and 
lower levels simultaneously. No orderly succession of events can be 

The final end of comprehension is to grasp the general significance 
and intention of the speech. But as far as I can see, we cannot assign 
the last place to this stage in all normal cases, for it cannot be called an 


abnormal case if the listener anticipates, before he has heard more than 
part of the speech, what the phrase-meaning is to be. That is or may 
be a case of rapid eduction. 

If we exclude eductions and supplementation as not belonging to 
comprehension in the narrower sense, we may of course say that the 
process begins with the perception of the words, and ends in the final 
comprehension of the speaker's ultimate intention. 

The variability in the process of understanding is no doubt of great 
value in making it possible for the hearer to adapt himself to unexpected 
meanings. If a word is used in a modified sense, the remainder of the 
sentence, whether coming before or after the word in question, will 
enable the hearer to correct his first interpretation, if it was incorrect. If 
the word has been interpreted at first according to its habitual meaning, 
the hearer is able without difficulty to modify his apprehension in the 
light of subsequent context. In this way the hearer is able to follow 
the precise shades of meaning of a speech, provided they are adequately 
expressed, whether they fall outside or inside the habitual semantic 
range of the words. The importance of this fact for sense-change is 
obvious. Modifications made by the speakers are at once communicated 
to the hearers. 

6.5. Intended and Comprehended Meaning. Identity and Discrepancy. 
The definition of meaning given above (3-27) makes meaning strictly 
individual: it is the subject's apprehension of the referent, and it 
varies from person to person. It follows that the meaning connected 
with a word may not be identical in the case of two persons who are 
conversing. I shall call the meaning given to the word by the speaker, 
intended meaning, and that given to the word by the hearer, 
comprehended meaning (Wellander, Studien I 9: gemeinte und erfasste 
Bedeutung) . 

The mental content as such is normally incommunicable. It must 
be translated into a sign system, and retranslated by the hearer into 
mental content. It is evident that the greater the similarity of the 
interlocutors in general knowledge and outlook, the easier and more 
perfect is their mutual comprehension. Considerable differences in 
these respects cause a correspondingly greater risk of misunderstand- 
ings (Paul, Prinz. 15, 78). 

Among the factors working for identity of intended and comprehended 


meaning, I mention first the range of the word, regarding which 
speaker and hearer have, normally, approximately similar notions (3.25). 
The hearer is debarred from attributing to the word meanings that fall 
outside the habitual range. 

Further, context, in the widest sense of the term, guides the hearer in 
making the specific objective reference intended by the speaker. I have 
shown above (3.23) that an identical objective reference guarantees 
mutual comprehension in so far as the two interlocutors are speaking of 
the same referent, however different their knowledge of and adjustment 
to this referent may be. Correct understanding hinges first and 
foremost on a correct objective reference. 

If the hearer makes an incorrect objective reference the utterance will 
in many cases not make sense; the mistake is then quickly detected 
and rectified. This is, I think, the most important censoring factor. 
However, mistakes due to misunderstandings and equivocation are 
deplorably common, even in scientific discussions. The speaker not 
seldom, although mostly unintentionally, makes use of ambiguous 
terms; for his own part, he intends one of the possible meanings, and 
he neglects to provide against the hearer's apprehending another. 

The difficulties of correct comprehension are in one way increased 
when we pass to phrase-meanings, since the hearer has not only to 
understand the single words, but to understand the phrase as a whole. 
I have described above the nature of this process. Since speech is 
discrete (5.51), the risk of misunderstandings is increased; it is not 
certain that the hearer will be able to make the correct eductions and 
supplementation (Biihler, Ber. 118 sqq.). On the other hand, the mu- 
tual determination of the words in a sentence lessen the risks for in- 
correct objective references (6.32). 

A meaning comprises also the speaker's subjective apprehension of 
the referent. We have therefore to ask, if and to what extent this 
element of meaning is communicated to the hearer, so as to enable him 
to evoke a subjective apprehension similar to that of the speaker, or at 
least to be informed of the speaker's apprehension even if he does not 
participate in it. I have already discussed this problem with regard to 
imagery (4.142) and emotional elements (4.153). Speaking generally, 
it seems probable to me (the problem has, as far as I know, not been 
properly investigated) that the subjective apprehension is communicated 
to the hearer in so far as it belongs to the habitual semantic range of 


the word as applied to the referent in question, but if it does not, the 
hearer has to educe as much as he can from the various factors of context, 
as described in the previous sections of this chapter. 

In this respect, then, there is always a discrepancy between intended 
and comprehended meaning; as long as the objective reference is 
correct, discrepancies with regard to the subjective apprehension may 
be unimportant for the practical purposes of speech, but as soon as 
we leave everyday life, and enter the realms of theoretical discussion, 
such discrepancies, too, may prove fatal to mutual comprehension. 
Every reader of this page will be able to provide instances from his 
own experience. 

An extreme case is complete non-comprehension, when the hearer 
either is unable to make the correct references for the individual words, 
that is to say, he does not understand them, or else understands the 
individual words, but fails to perform the further necessary operations 
of eduction and supplementation (cf. Otis' instance, 6.37). The reason 
may be insufficient acquaintance with the matter, i. e. inability to 
integrate the referent in any system of relations known to him. 
(Ward 301; cf. Thorndike, Reading 323 sqq., and Spearman 122 on 
failure in execution). In general, if the speaker employs expressions 
chosen with regard to his own mentality only, they may not call forth 
adequate references in the hearer's mind (Cf. Husserl II. i. 418, Marty, 
Unt. 463 — 464, with additional instances, Delacroix 440). 

Another extreme case is the rejection by the hearer of a communica- 
tion as not compatible with facts. The hearer is then able to make the 
required eductions and supplementation, but he is not able to make a 
reference to facts, since in his opinion there are no facts to which the 
statement can be referred. (Cf. Marty, Unt. 362; and 5.52). 

Goteb. Hogsk. Arsskr. XXXVIII : i 


7.1. Definition and Preliminary Delimitation. 

7.11. Change and Stability. Every utterance is the expression of 
the momentary state of mind and purposes of the speaker. Every 
man has — within reasonable Hmits — thoughts and feelings to ex- 
press that are peculiarly his own, and also his own way of expressing 
them. Moreover, many referents are constantly changing, compelling 
changes in the utterances by which they are symbolized. If speech 
were determined only by these factors, it would vary indefinitely. 

But every utterance has also to symbolize stable referents, and to 
carry a communication, and these functions work for stability. They 
compel the individual speaker to conform — again, within reasonable 
limits — to the ruling language system (cf. 2.11), otherwise he would 
not be understood by his hearers. A limit is thus set to the variations 
of speech. 

An utterance is, on the one hand, the product of the momentary 
situation and the individual speaker's reaction to it: as such it is 
subject to variations in different respects; on the other hand, it is a 
realization in concrete linguistic material — words, meanings, and 
sentence schemes — of a portion of the language system: as such it 
tends to conformity with previous utterances referring to the same 
kind of referents, and spoken by the same or other speakers. The 
language system serves as a norm for the individual utterance, 
moulding it in the traditional form. 

The actual course of semantic development emerges as the result of 
these conflicting tendencies; and according as one or the other of them 
prevails with regard to a complex of word and meaning, we find that 
the complex remains stable throughout long periods, or, conversely, 
that it offers a picture of constant change. 

Change and stability are equally normal phases in the history of 
language; only the total absence of either would be abnormal. 


7.12. Definition of Sense-change {Change of Meaning). Wlien a 
word is employed to express a meaning which it has not previously 
expressed, we have, from the point of view of the word, a change of 
meaning. Often a change of meaning is also a change of referent (for 
details, see the following chapters); from the point of view of the new 
referent, we have then a change of name. The former point of view 
is 'that of semasiology, the latter that of onomasiology (cf. 1.2). 

If a word is used with a new meaning once only, the matter is of no 
importance for Unguistic development; but if it is thus used repeatedly, 
and by a comparatively large number of speakers, the incidental use 
becoming habitual, there is established a permanent connection be- 
tween the word and the new meaning. The traditional semantic range 
of the word is modified accordingly, often the referential range also. 

A change of meaning is not necessarily a change of referent. Accord- 
ing to the definition adopted in the present work (cf. 3.27) a change 
in the manner of apprehending a referent is also a change of meaning. 
The word then symbolizes a referent or some referents within its tra- 
ditional referential range, but it expresses a changed subjective appre- 
hension of them. This type of change is especially instanced by ade- 
quation (ch. 14). 

I define change of meaning as the habitual modification, among a 
comparatively large number of speakers, of the traditional semantic 
range of the word, which results from the use of the word (i) to denote 
one or more referents which it has not previously denoted, or (2) to 
express a novel manner of apprehending one or more of its referents. 

7.13. Change and Fluctuation. The wording of the definition in 
the preceding paragraph brings us immediately to the problem of 
change and fluctuation, and the distinction between them. 

It has been pointed out in the previous chapters that, strictly speak- 
ing, the psychic processes always vary from instance to instance, and 
from individual to individual. The variations are largely variations 
of context — of "setting" — due to the constantly shifting circum- 
stances in which the word is employed, and to the similarly shifting 
apprehension and purposes of the speaker. 

I have referred above (3.24) to the circumstance that "what for one 
is no more than a kodak, develops for the perception of another into 
a reflex, extension, swing-front, focal-plane, anastigmatic hand-camera 
de luxe", and I have stated that I do not consider that the word kodak 


has the same meaning for the two individuals. But the ignorance of 
the one or the special knowledge of the other does not constitute a 
permanent change of meaning for the word. They apprehend the 
referent differently, but these differences are a matter of individuals, 
and, at least for the moment, not a matter of the speaking community 
or of any considerable group within it. They have not yet led to any 
permanent modification of the traditional semantic range of the word 
kodak. I therefore regard the differences as fluctuations, not as changes 
of meaning. 

Nor do I regard as changes of meaning the shifts of apprehension 
involved in occasional speciaHzation (cf. 4.24), or particularization 
(4.23). If these processes become habitual, a sense-change may result 
(cf. 14.8). 

The expression "the community or any considerable group within 
it" covers another point of uncertainty. There are, obviously, all 
degrees between the individual variation and the change adopted by 
the whole speaking community. Shifts peculiar to individual trades 
and professions are generally recognized as definite changes of mean- 
ing, even if the trade or profession in question should happen to be 
exercised only by a comparatively small number of people. On the 
other hand, there are modifications of meaning which are known only 
to a family, a coterie of friends, among the staff of an institution, or 
the members of a school class, or any other similar Httle group. These 
modifications of meaning must be classed as fluctuations as long as 
they do not pass the boundaries of the group where they originated. 

A third point of contact for fluctuation and change Hes in the factual 
variations of the referent. There are, no doubt, constant alterations in 
the manufacture and construction of kodaks, as improved types are 
produced, but these variations cannot be said to lead to a sense- 
change until their cumulated effect, working in one direction, results 
in the production of something that presents itself as a new sub-cate- 
gory of kodaks, apprehended as such at least by some considerable 
group of speakers. Cf. also 12.13. 

It will be evident from these remarks that the discrimination of 
fluctuation and change of meaning is a problem of great difficulty, 
which requires much further investigation. In practice it will often 
be impossible to assign an individual instance definitely to one or 
the other class. We have to be content, at present, with formulating 


the rule that a change of meaning must involve a habitual modification 
of the traditional semantic range of a word among a comparatively 
large group of speakers. In other words, the change must have be- 
come incorporated in the language system of the group. 

7.2. The Seven Classes of Sense-change. 

7.21. The Fundamental Point of View. If we compare a new 
meaning of a word with the earlier meaning out of which it arose, we 
find that the change has caused a modification of the mental content 
that constitutes the meaning. The logical relation between the earlier 
and the later meaning has been utilized by many writers to classify 
the semantic changes. It has been found, however, that such a classifi- 
cation cannot be made exhaustive (cf. 1.3). 

We may, further, study the psychic processes which result in, or are 
involved in, the changes of meaning, and classify the changes on that 
basis. Wundt and other writers have proposed psychological classi- 
fications, but their systems have not been satisfactory, chiefly owing 
to the fact that they were founded on an inadequate analysis of the 
nature of meaning, and on an insufficient linguistic material. 

For my part, I began by analj^sing and sorting historical instances 
of sense-change, mainly with regard to the psychic processes involved. 
By this empirical method I arrived at the result that there are seven 
main classes of change; they are enumerated in the following paragraphs. 
I then turned the matter round, in order to ascertain if these classes 
were statistical types only, or if they formed an organic and coherent 
system. I found that the latter was so to a certain extent. By ap- 
plying the psychological principles explained in the preceding chapters, 
and the definition of meaning proposed in ch. 3, the seven classes can 
be arranged in a satisfactory system. The principles are explained 
in 7.3, and in detail for each class in the relevant chapter. 

The adoption of psychological principles for the distinction of the 
main classes does not imply that the logical points of view are to be 
discarded altogether. It will appear from the detailed treatment below 
that they are indispensable for making further divisions in each class. 
It is a mistake to assume that a satisfactory semantic classification 
can be founded on psychological considerations alone. These give us, 
to begin with, onty a limited number of main classes or groups, too 
large for practical purposes; and logical points of view must be ap- 


plied for breaking them up into smaller groups of a more convenient 
size. Secondly, many psychic processes may lead, indiscriminately, 
to one or the other semantic result; conversely, different processes 
may give the same result. Since it is sense-changes I am classing, such 
processes are useless for my purposes. I can make use only of phenom- 
ena that are constant concomitants of a group of changes, and of 
that group alone. Moreover, we cannot always determine with pre- 
cision what psychic processes are involved in one or the other change 
of meaning. 

It should be added that the logical relations holding between mean- 
ings are often direct reflections of the factual relations between the 
corresponding referents, and the latter relations are of great import- 
ance for the changes affecting their names. I refer for details to the 
following chapters. 

In the present chapter, the general principles of sense-changfe will 
be the topic of discussion. In order to provide a material to work on, 
I shall first give a typical instance of each of the seven classes, with 
a brief analysis of its main characteristics. 

7.22. Class I. Substitution. Substitutions are sense-changes due 
to external, non-linguistic causes. The word sAi/), at present, may have 
meanings that were unknown at a time when steamships, motor-ships , 
airships, etc., were not yet invented; and it will no doubt go on gathering 
new meanings in future, as new types of ships are built. The new 
referents were apprehended by English speakers as belonging to the 
category of ships, and they were therefore denoted by the same name. 
To travel nowadays calls up the thought of trains, motor-cars, steam- 
ships, and air-planes, while a hundred years ago it made people think 
of horses, stage-coaches, and sailing-ships. The cause of such shifts, 
as well as others of cognate types, is the development of technique 
and other cultural factors, which lie altogether outside language and the 
speech activity. Language only registers the change. In this respect 
substitutions differ from all the other types, which without exception 
are due to psychic causes connected with the speech activity. 

7.23. Class II. Analogy. Analogy plays as important a part in 
the semantic system as it does in the morphological system. There 
are several types, of which I shall quote one. The English adjective 
fast has two meanings that are almost contradictory, 'firm, immovable', 
and 'quick'. There are no intermediate senses that might have served 


as links in a development from the former to the latter. The adverb 
fast (ME. faste), on the other hand, shows a continuous development 
from the earlier sense 'firmly, immovably' to the later sense 'quickly'. 
It is evident that, as stated by the NED, the adjective has "borrowed" 
the sense 'quick'. In other words, when the adverb had acquired the 
new sense, it was, by analogy, extended also to the adjective; our 
linguistic feeling is accustomed to adjectives and adverbs of the same 
stem having strictly correlated senses. 

7.24. Class III. Shortening. If, for some reason, a word is 
omitted from a compound expression, which still retains its meaning, 
the remaining words or word have to carry the total meaning that 
formerly belonged to the whole expression. If the omission becomes 
habitual, the result may be a sense-change for the remaining word or 
words. Thus, private 'common soldier' is a shortening of private sol- 
dier; the noun has been omitted and the adjective has acquired nom- 
inal character, as shown by its ability to take a plural ending. 

7.25. Class IV. Nomination. All speech is more or less intentional, 
in so far as it is an instrument for the promotion of purposes. But 
the intentional character of speech concerns primarily its import, not 
its form. The selection of words, as well as their syntactical arrange- 
ment according to traditional sentence-schemes, is left to the care of 
lower centres, the activities of which are almost completely automatic, 
and we intervene consciously only when something goes wrong and 
has to be rectified. However, sometimes we also pay attention to the 
form of our utterance. We wish not only to present the topic to the 
listener in an objectively correct way, we wish also to make the hearer 
take up a definite attitude towards it, to perceive it in a certain colour, 
and so on. We then strive to select words suitable for the purpose, and 
it is this intentional selection of words that is here meant by the term 
intentional (cf. 11. 13 for details). In such cases the speaker may find 
that he cannot make his point without emplo3dng one or more words 
in a new way. When Keats begins his Ode on a Grecian Urn with the 

Thou still unravished bride of quietness, 
Thou foster-child of silence and slow time — 

the metaphors are intentionally chosen for their power of expression 
and impression, and they are new in this use. Nominations are trans 

1 68 GUSTAF STERN 7.25 

fers in which a name is intentionally transferred from one referent to 
another. The distinction between intentional and unintentional is 
often vague. 

7.26. Class V. {Regular) Transfer. I include in this class only 
unintentional transfers, based on some similarity between the original 
(primary) referent of the word and the new (secondary) referent. The 
condition of the transfer is the speaker's perception of the similarity, on 
which he bases a transfer of the name from one referent to the other. 
Thus, thin objects resembling a leaf in shape, may be called leaves. 
In other cases, the basis of transfer may be a similarity of function, as 
when bed is used for the foundation of a steam-engine or other machine, 
because it supports the machine as a bed supports a person lying on it. 

7.27. Class VI. Permutation. In the phrase he is counting his 
beads, the last word (MB. bedes) originally meant 'praj'ers'. In the 
Middle Ages, prayers were mostly Pater Noster and Ave Maria, which 
were said repeatedly, being counted by means of the little balls on a 
rosary. To count one's prayers and to count the balls of one's rosary 
was, then, almost the same thing as regards the purpose of the 
action. A person saying he is counting his beads, which meant 'he is 
counting his prayers', would in reality see the man referred to counting 
the balls of his rosary. There was thus set up a close association between 
the word beads, with its primary meaning 'prayers', and the notion 
of 'balls'. Moreover, it was often quite immaterial whether the phrase 
was understood in the one way or the other; it remained, nevertheless, 
an adequate designation for its total (phrase) referent. The meaning 
is "equivocal" (cf. 13.12) but either interpretation will serve. The 
result is that, finally, the word beads is employed to designate the 

7.28. Class VII. Adequation. I have taken the term adequation 
from Stocklein, who describes some types belonging to this class. It 
is, essentially, an adaptation of the meaning to the actual characteristics 
of the referents which the word is employed to denote, but in contra- 
distinction to substitution, the cause of the shift lies in the subjective 
apprehension of the speakers. Horn was originally 'an animal's horn' 
and was so called even when used for producing signals or music. The 
chief characteristic of such a horn was after all not the material from 
which it was manufactured, but the purpose for which it was used; 
and the notion of this purpose became the predominant element of 


the meaning of horn — when applied to such objects. The meaning 
of the word was adequated (adjusted) to the real characteristics of 
the referents. When the adequation was completed, the word could 
be transferred to similar musical instruments manufactured from other 
material. The condition for the transfer was the preceding adequation 
of meaning. 

7.3. General Causes and Conditions o£ Sense-change. 

7.31. The Three Immediate Causes of Sense-change. After having 
established empirically the existence of seven main classes of sense- 
change, it is next necessary to ascertain if these classes are statistical 
types only, or if it is possible to formulate a theory through which they 
can be showed to form an organic system. The point of view will 
obviously have to be genetic, i. e. a consideration of the processes 
involved in or causing the changes. A deductive argumentation should 
show the theoretical possibilities, and these should agree with the 
results of the empirical investigation. 

I think we may find such a point of view in the definition of meaning 
formulated in ch. 3. I have stated there that an actual meaning is 
adequately determined by three factors, the objective reference, the 
subjective apprehension and the traditional range; in other words, by 
its relations to the referent, the subject (speaker or hearer), and the 
word. I shall call them the referential, the subjective, and the verbal, 
relations. I further stated that if two only of these factors are stable, 
while the third varies, the meaning varies with the third factor. No 
fourth factor could be found in verbal meaning. 

If this is correct, it must follow that any sense-change has as its 
immediate cause a change in one of the three relations mentioned.^) 
As far as I can see, this agrees with the facts. 

Analogies and shortenings are, primarily, a modification of the verbal 
relation; substitutions, nominations, and transfers, a modification of 
the referential relation; permutations and adequations, a modification of 
the subjective relation. For details, I refer to the following chapters. 

7.32. Further Causes: External and Linguistic. The three immediate 
causes of sense-change must, in their turn, be evoked by other factoxs, 

^) It might be more correct to say that every sense-change consists primarily 
in a shift of one of the three relations. It is partly a difference in the point of 
view, and I do not think that the terms I employ will lead to misunderstanding. 

170 GUSTAF STERN 7.32 

and the latter may furnish further points of view for more fundamental 
distinctions. To a certain extent this is the case. 

The instances given above have shown that substitutions are caused 
by external, non-Hnguistic factors, while all the other six classes are 
caused by linguistic and psychic factors. This is evidently an essential 
distinction from the point of view of genetic classification. The 
changes of a referent that are reflected in substitutions depend alto- 
gether on factors outside language and the speech activity. Language 
only registers the changes. The linguistic causes of change, on the other 
hand, are psychic processes occurring in conjunction with the speech 
activity, in the production or comprehension of speech. 

I make a first distinction between substitutions, on the one hand, 
and the other six classes, on the other. 

7.33. Further Causes: Intentional and Unintentional Processes. 
There is one other important distinction applicable to the whole field 
of sense-change: the distinction between intentional and unintentional 
processes (cf. 7.25 and 11. 13). It must evidently make a great 
difference in several respects if the changes are due to the automatic 
action of spontaneous psychic processes, or if the speaker's will inter- 
venes. In the unintentional changes, only the simplest psychic processes 
can be involved, which start spontaneously from some initial impulse 
and pass off unperceived. It shotdd therefore be possible to distinguish 
classes of change corresponding to the types of unintentional processes 
which are possible. In the case of intentional changes, on the o'^er 
hand, an arbitrary factor is introduced which may cut across all ordinary 
psychic combinations, obeying the dictates of momentary and purely 
individual impulses.^) 

For my purposes, this is clearly an essential distinction; and it is 
applicable to the whole field. 

We have now two points of view for changes due to linguistic causes. 
One is the tripartition based on the immediate causes of the shift, as 

1) "Erklaren ist 'klassifizieren' und 'ordnen' und sonst kann es nichts sein. Die 
'Klassifizierung' verspricht leichtere Erfolge als die 'Ordnung', die nur auf natur- 
wissenschaftlichem Gebiete ihrer Stabilitat wegen zu sicherem Urteil fiihrt, 
wahrend die Labilitat des bewussten psychischen Geschehens wegen ihrer Ein- 
maligkeit nur zu moglichem, bzw. wahrscheinlichem Urteil fiihrt. Doch lehrt. 
diese Unterscheidung als Grundsatz: wo Stabilitat in der Beziehung Ursache- 
Wirkung zu finden ist, ist kern Bewusstsein. Wo Labilitat, also die Beziehung 
Grund-Folge zu finden ist, ist Bewusstsein. Anders ausgedriickt: Um so geringer 


involving a modification of the verbal, the referential or the subjective 
relation. The other is the bisection into intentional, and unintentional 
changes. Which of these two principles should be applied in the first 

A comparison between them shows, first, that analogies and 
shortenings may be either intentional or unintentional (cf. chs. 9 and 
10). It would, at least in this case, not be expedient to make that 
distinction our first point of view, since it would necessitate the 
splitting up of two clearly defined classes. 

Taking next the changes due to a modification of the subjective 
relation, we find that permutations and adequations are always 

But the case is different with regard to shifts caused by a change of 
the referential relation. They embrace a very large number of smaller 
groups, ranging from the metaphors, as quoted in 7.25, to regular 
transfers of the simplest kind. In this mass of changes, I can see no 
other possible line of division than the distinction intentional/uninten- 

Consequently I make the three immediate causes of sense-change 
my first consideration, dividing the six classes II — VII into three 
groups. To one of these only I apply the distinction between intentional 
and unintentional. 

Within the other two groups, the distinction of classes must be em- 
pirical. We find that the verbal relation may be modified through 
analogy or shortening, and that the subjective relation may be mo- 
dified in regard to a phrase referent or to a word referent. Accordingly, 
each group is divided into two classes. 

7.34. The Ultimate Causes of Sense-change. Going still further 
back in the causal chain, we have to ask what factors release the 
processes, intentional or unintentional, that lead to modifications of 
the three relations of meaning, and thus to sense-changes. The 

das Bewusstsein, urn so geringer sind vSchwankungen in der Beziehung Grund- 
Folge und viceversa". ly. Jordan, Idg. Forsch. 44 (1927) 85. — "Un pur 
mecanisme qui se deroule automatiquement n'est sujet qu'a des dereglements, 
a des deraillements, a des achoppements provenant du mauvais etat de telle 
ou telle de ses pieces ou de I'insuffisance de Tajustement . . . Un mecanisme 
qui se deroule sous une surveillance est expose en outre aux f antes du surveillant" 
(Delacroix 163). 

172 GUSTAF STERN 7.34 

discussion is first confined to the shifts due to linguistic causes (classes 
II — VII), and we begin by separating the last two, permutation and 
adequation, from the first four. 

With regard to the latter ■ — analog}^ shortening, nomination, and 
transfer — the ultimate causes that release the process are, generally, 
the functions of speech. I have stated above (5.4) that a speaker 
adjusts himself, and if necessary also his instrument, speech, to the 
purposes for which he employs it. If the actual resources of the instru- 
ment, or, more precisely, the momentarj^ resources of the speaker, 
his personal language system, are not equal to the task set to it, the 
speaker may modify it in various ways: thus he may use words with 
a new meaning.^) All changes of these four types are effected by the 
speaker; the hearer's part is only to understand. 

Taking the four functions of speech one by one, the speaker may 
employ words in a new way, (i) in order to communicate more clearly 
and adequately to the hearer the import of the utterance (the com- 
municative function); (2) in order to symbolize the referents more 
adequately (the symbolic function); (3) in order to express his thoughts 
and feelings more adequately (the expressive function); or (4) in order 
to make a stronger effect on the hearer (the purposive function, which 
more or less involves the other three). 

These statements are obviously applicable to the intentional changes, 
but also, in a slightly different way, to the unintentional ones. The 
apprehension of similarity underlying a shift like that of leaf or bed 
(7.26) presents itself spontaneously to the speaker's mind; and the 
similarity has, in itself, nothing to do v/ith the functions of speech. 
But a speaker does not accept and make use of all associations that 
arise; his acceptance and employment of a spontaneous association as 
the basis of a fresh designation of a referent, must be assumed to be 
conditioned by the requirements of the speech functions. Even if 
individual speakers make use of unsuitable names, this does not con- 
stitute a real change of meaning (cf. 7.13); we must assume that the 
new name is not accepted by the speaking community or any con- 
siderable group within it, if it does not in some way meet the demands 
which are made on speech as an instrument for the promotion of 
purposes. I assume that, in the same way, analogies and possibilities 
of shortening that offer themselves automatically to the speaker are 

1) Cf. Stern, Psych. 129, on "Sprachnot". 


accepted or rejected according to their suitability for the purposes of 

We may, then, say of unintentional changes as well as of the 
intentional ones, that they are the result of the striving of speakers to 
adapt speech to the purposes for which they make use of it. 

Turning now to classes VI and VII, permutation and adequation, we 
note an important difference. In the four classes previously discussed, 
the change is effected by the speaker, and "in one movement": the 
hearer's part is only to understand. In the last two classes the 
circumstances are different. The change of meaning from 'prayers' 
to 'little balls' is preceded by, and has to be prepared for by, repeated 
associations between the word beads and the notion of little balls used 
for counting prayers. It is quite possible that, during the period of 
preparation or transition, the speaker may intend the word to be 
understood in the earlier sense, while the hearer actually interprets 
it in the new way. This does not, as it would do in classes II — V, cause 
an ambiguity and a misunderstanding. Similarly with adequations; 
during the period of transition, a man speaking of horns 'musical 
instruments' may be thinking of them also as being made of animals' 
horns, while for the hearer this element of meaning has disappeared; 
and vice versa. 

As in other unintentional changes, the association arises spontan- 
eously, as a result of the unconscious psychic activities, and I assume 
that the possibility of a change is made use of by the speaking com- 
munity or group only when it serves some purposes connected with 
the functions of speech. 

This term should be taken as embracing aesthetic pleasure as well 
as effect, agreement with a prevailing fashion, and any other factor 
that may please a speaker. 

With regard, finally, to substitutions, I have stated that they are 
due to external, non-linguistic causes. But it is undeniable that we 
use the same name for a new ship which is different from other ships, 
but also in important respects similar to them, because we want a 
name for it, that is to say, in order to enable speech to fulfil its 
symbolic and communicative functions in regard to the new referent. 

I conclude, then, that most sense-changes are the result of the striving 
of speakers to adjust speech yet more closely to the functions which 
it has to perform. 

174 GUSTAF STERN 7.34 

One other factor of general character should be mentioned: the 
economic tendency (Bequemlichkeitstrieh). It has, apparently, more 
influence on the development of sounds and forms than on that of 
meanings, but it is not altogether without importance for the latter. 
It is probably responsible for some shortenings (cf. 10.185), and it does 
not seem unreasonable to assume that the tendency to employ words 
of more general import, which are often recalled more easily than a 
precise term, is sometimes due to the choice of the easier alternative 
(cf. ch. 14). 

The economic tendency would seem to be sometimes in conflict with 
the striving to adapt speech to its functions. Our knowledge of the 
matter is at present very imperfect, and I have to leave it with these 

We may note, too, that the ultimate causes of sense-change, the 
functions of speech, cannot be utilized as a basis of classification, since 
any one of them may lead to any one type of sense-change. Even an 
elliptical phrase may be clearer and more adequate, because less 
cumbrous, than a more complete utterance, and it is often superior 
as a means of expression or impression.^) 

7.35. The General Conditions of Sense-change. I have stated 
above (7.1 1) that the symbolic and communicative functions of speech 
compel the speaker to conform more or less to the ruling language 
system, in order to be understood by his hearers, and that this 
circumstance sets a limit to the possibilities of sense-change. The 
analysis of the process of comprehension in ch. 6 has shown that the 
hearer is able, by the help of context, to interpret new meanings 
correctly, en fonction du schema d'ensemble, even if they should deviate 
considerably from the traditional language system. The speaker has 
therefore a fairly wide latitude for his tendency to variation. He may, 
within wide limits, shift the meaning of his words to suit his momentary 
purposes. But there are limits, and it is obviously the exigencies of 
symbolization and communication that play the main part in imposing 

Closely connected with this factor is the tendency to conformity to 

^) I may remark here that Paul was evidently right in making "die gewohnliche 
Sprechtatigkeit" the cause of all sense-changes. Only he was not, at that time, 
able to show what elements and processes constitute this extremely complex 


the ruling habits of speech. As the following chapters will show, the 
changes of meaning are confined to a comparatively small number of 
types, with which we are familiar. I suggest that this fact, too, is due 
to an adjustment to the hearer's possibilities of comprehension; the 
common types are those that make no difficulties; difficult types are 
more rarely resorted to, and habits are thus estabUshed which, in their 
turn, help to mould new changes to old and familiar patterns. 

We have at present no detailed knowledge of these problems. For 
various special conditions I refer to the following chapters. 

7.36. Scheme of Classification, and Final Remarks. The statements 
above concerning the general causes and conditions of sense-change 
have shown that the latter cannot provide any points of view for the 
classification, but the former enable us to group the seven main classes 
in a coherent scheme. As mentioned, I take for my first basis the dis- 
tinction between external and linguistic causes of change. The six 
classes falling under the latter heading are grouped according to the na- 
ture of the primary causes of change. 

This gives us the following scheme: 

A. External Causes Class I. Substitution. 

B. lyinguistic Causes: 

I. Shift of Verbal Relation a. Class II. Analogy. 

b. Class III. Shortening. 

II. Shift of Referential Relation a. Class IV. Nomination. 

b. Class V. Transfer. 

III. Shift of Subjective Relation a. Class VI. Permutation. 

b. Class VII. Adequation. 

The theoretical principles gained from typical instances of each class 
are fairly clear, although in some cases there are difficulties in formula- 
ting precise definitions. The theoretical formulation has, subsequently, 
to be employed in classing less typical material, and it is here that the 
greatest difficulties are encountered. Very often we do not know the 
history of words with sufficient accuracy to assign them with certainty 
to one or the other class. A still more serious difficulty is, however, 
the fact that two, or even more, different processes may give exactly 
the same morphological and semantic result. In these cases — it is 
generally a question of specific groups — we have to content ourselves 
with registering the alternative explanations. As far as I can see, a 

176 GUSTAF STERN 7.36 

certain amount of such overlapping is unavoidable, but the classes 
should be so chosen and defined that the central groups of t5^ical 
instances are not affected by the ambiguity. There are also inter- 
mediary groups, the placing of which is doubtful. This is a draw- 
back attendant on all linguistic classification based on meanings; I 
need only refer to the many thorny problems of syntactical discrimina- 

No systematic knowledge of a historical material is possible without 
distinguishing types of development. Since the development of language 
follows its own laws, logical or psychological points of view are 
relevant only in so far as they serve to explain the factual linguistic 
phenomena. These are the main thing. The question to be asked is 
always: what has happened to the word and its meaning during the 
course of development? 

The groups to be distinguished are the groups that show in the ma- 
terial itself. If they correspond to definite psychic processes, or to de- 
finite logical relations between the meanings — reflecting factual rela- 
tions between the referents — then these psychic processes and logical 
relations should be employed for purposes of definition and analysis. 
But if a psychic process or a logical relation — however important in 
itself — does not correspond to a discernible group in the linguistic ma- 
terial, it is useless for classifying purposes.^) Failure to recognize 
the latter principle has been the cause of many attempts to press 
the linguistic material into moulds of foreign origin, into which it does 
not and cannot fit. 

Every scientific classification is, or should be, a method of analysis, 
and should, consequently, adapt itself as closely as possible to the es- 
sential characteristics of the facts analysed. 

7.4. Some Relevant Problems. 
7.41. The Spread of New Meanings. Sense-changes are normal 
phases of the linguistic development. It follows as a corollary that a 
sense-change is adequately explained if we can explain its happening, 
as a normal process, in the mind of one person. It is evident that 
the processes leading to a sense-change, like any normal psychic events, 
can (i) occur repeatedly in the mind of the originator, or (2) be imitated 

^) The analysis of such processes and relations may be important and interesting 
from other points of view. 


by other speakers to whom the new use of the word will present itself 
as a normal way of using speech, or (3) happen, independently of imita- 
tion, in the minds of other speakers, who find themselves confronted 
by the same linguistic task as the first. 

The last two factors will contribute to the spread of new meanings in 
the speaking community. For unintentional changes, both methods 
seem equally possible and probable; for intentional changes, which 
may be due to a "singular" combination of circumstances that is not 
likely to be repeated, method no. 2 is perhaps more common. 

The problem for the present investigation is the origin of new mean- 
ings, not their dissemination, but a few remarks may be offered on 
the latter question. 

The question why some new meanings catch on and are universally 
employed, while others have only an ephemeral life is one of the most 
difficult problems in semasiology. We can, in general, say that the 
interest directed towards anything is the main reason for its being 
recollected and re-evoked.^) 

In accordance with the dynamic point of view adopted, I assume 
that the main factor of interest about an utterance — and its consti- 
tuent words and meanings — is its successful functioning. It has 
been demonstrated with regard to children's speech that the pleas- 
ant effect of an utterance is the chief fixating factor, leading to 
recollection and repetition of the phrase.^) I think we should as- 
sume that words in ordinary speech are chosen mainly for their effect 
in respect of the various functions of speech, as adapted to the mo- 
mentary purposes of the speaker. The rapid success and equally rapid 

^) "It is equally clear how important for the recall of the sensory excitations 
in the future, and their entry into the intellectual capital of the individual, must 
be their success in rousing some interest. Those, on the contrary, which succeed 
physiologically in arriving at their normal point, but remain unconscious because 
they are not associated with any affectivity whatever, should be considered, if 
not all, at any rate for the most part, as lost for the intellect; for they would 
with great difficulty find occasion to be evoked before their over long "inaction" 
caused them to disappear also as mere mnemonic accumulations". Rignano 68. 

^) Cf. Freudenthal 47 sqq., esp. 57, with quotations from other writers. Also 
Ivindworsky, Zs. f. Psych. 92, 1923, 369, in a review of Selz: "Nur das erfolgreiche 
Verfahren ist gleichzeitig mit dem Ziel selbst im Bewusstsein"; and Helson, Am. 
J. of Psych. 37, 48 on the stamping-in effect of the pleasant consummation, with 
quotations from Koffka, Psych. Entwicklung; also Pos 126 sqq. 

Goteh. Hossk. Arsskr. XXXVIII : i 


disappearance of fashionable catchwords are thus explained: as long 
as they are fresh and amusing, everybody "in the swim" will use them, 
only to drop them when people begin to tire of the incessant repetition. 

The principle seems incontestable, but its application to the facts of 
language bristles with difficulties, and is broken through by numerous 
exceptions and cross-influences. For instance, in OE the common 
word for 'throw' is weorpan. Why is weorpan, in this use, succeeded 
by cast, and cast in its turn by throw? Cases like these, and they 
are innumerable, would seem to be inaccessible to our present 
methods, and at the present stage of knowledge. ^ 

The earlier view of language as mainly — or only — a means of com- 
munication led to the opinion that linguistic development is due essen- 
tially to the two conflicting tendencies to ease and to clearness [Be- 
quemlichkeitstrieh und Deutlichkeitstrieh): the best formulation is that 
which is easiest to find and easiest to understand (det som lattast givet 
lattast forstas. E. Tegner, Svensk Tidskrift 1874, p. 130). But it is 
evident that this formula altogether misses the function of speech for 
symbolization, expression and impression, and that it is therefore 
inadequate. (Cf. Noreen, Spridda Studier I 143 sqq). 

The tendency to clearness is covered by the general formula that I 
have given, but the tendency to ease w^ould appear to be an addition to 
it (cf. 7.34). 

The closely related problem concerning the disappearance of mean- 
ings is equally, or more, obscure. We can discern some typical groups, 
but we do not know what portion of the total field is covered by these 
groups (Cf. Sandfeld, Sprogv. 126 sqq., Noreen, Spridda Studier II 126 
sqq., Holthausen, GRM 7, 184 sqq., Carnoy 28, quoting Gillieron. 
Additional instances in Gillieron, Pathologic et Therapeutique verbal). 

7.42. The Transmission to New Generations. The influence of the 
child, or more precisely, the discontinuity of the linguistic tradition, on 
phonetic change, has been the subject of much discussion (cf. Jes- 
persen, Language 103 sqq., especially 172 sqq., with a review of the 
relevant facts, and Delacroix 179 sqq. with a criticism of some 
current opinions). 

It does not seem unreasonable to assume that the discontinuity of 
transmission due to the entrance of new generations into the speaking 
community, should be responsible also for some semantic changes. 
What a child hears and learns is actual speech, on the basis of which he 


forms his own language system. Discrepancies between the meanings 
intended by adult speakers, and the meanings actually apprehended 
by the child are often possible. Paul thinks that we have to assume a 
type of sen^e-change "die darauf beruht dass der fiir die altere Genera- 
tion usuellen, Bedeutung von der jiingeren eine nur partiell damit iiber- 
einstimmende untergeschoben wird. Das Gebiet dieser Art des Wan- 
dels werden wir aber auf die selteneren und nicht leicht klar zu fixieren- 
den Begriffe einzuschranken haben, da bei anderen die allmahliche 
Korrektur nach dem bestehenden Usus nicht ausbleiben kann" (Paul, 
Prinz. 86). The shift is due to the possibility of the younger generation 
misunderstanding a metaphorical or transferred sense as being the 
chief sense of the word. For instance, they may happen to hear Fuchs, 
'fox' for the first time in the sense of 'bay horse', or Kamel in the 
sense of 'stupid man'. 

Paul's instances are unfortunately chosen, since there is no tendency 
in standard German to make the secondary meanings of Fuchs and 
Kamel the main senses of these words, discarding the original 
meanings. The words are consequently instances of the fact that, even 
if children should misunderstand them, they will be corrected to con- 
formity with the current usage, and the language system will remain 

Jespersen (lyanguage 175) speaks of changes "that have come about 
with a leap, and in which it is impossible to find intermediate stages 
between seemingly heterogeneous meanings, as when head, from 
meaning 'a prayer', comes to mean 'a perforated ball of glass or amber'. 
In these cases the change is occasioned by certain connections, where 
the whole sense can only be taken in one way, but the syntactical 
construction admits of various interpretations,^) so that an ambiguity 
at one point gives occasion for a new conception of the meaning of the 
word. The phrase to count your heads originally meant 'to count 
your prayers', but because the prayers were reckoned by little 
balls, the word heads came to be transferred to these objects, and 
lost its original sense. It seems clear that this misapprehension could 
not take place in the brains of those who had already associated the 

^) A very clumsy expression! Of course it is the whole sense that can be taken 
in two ways, but the referent may remain the same. The syntactical construction 
has nothing to do with the sense-change. I refer to the detailed analysis in 13.11, 
which shows clearly enough that there are intermediate stages. 

l80 GUSTAF STERN 7.42 

word with the original signification/) while it was quite natural on the 
part of children who heard and understood the phrase as a whole,^) but 
unconsciously analysed it differently from the previous generation." 
(Cf. Stern, Kindersprache 307 sqq.). 

Meillet has similar opinions. The earlier sense of a word, which is 
still predominant for an older generation, may be effaced in the appre- 
hension of the younger generation. Thus the word saoul, originally 
meaning 'rassasie', was applied euphemistically to people who were 
rassasies de boisson 'replete with drink'. The child on hearing 
such an expression wotdd be likely to associate with it the simple 
notion of 'drunk', and in this way saoul has become the habitual 
word for that notion, without any euphemistic force. "Cette discon- 
tinuite de la transmission du langage ne suffirait a elle seule a rien ex- 
pliquer, mais, sans elle, toutes les causes de changement auraient sans 
doute ete impuissantes a transformer le sens des mots aussi radicalement 
qu'il I'a ete dans un grand nombre de cas: d'une maniere generale 
d'ailleurs, la discontinuite de la transmission est la condition premiere 
qui determine la possibilite et les modalites de tous les changements 
linguistiques" (Meillet, Ling. 235 — 236. Similar opinions are ex- 
pressed by Jaberg, Zs. f. rom. Phil. 27, 30). Meillet thus attributes 
to this factor a much greater importance than does Paul. 

Arguments of this kind are founded on two assumptions, not always 
explicitly stated, and, in my opinion, both of them erroneous. The 
first is the assumption that in every "normal" sense-change there must 
be a close logical affinity between the primary and the secondary 
meanings, otherwise the change is not normally explicable, but extra- 
ordinary factors must be adduced to explain it, as, for instance, childish 
misunderstandings. It will be abundantly evident from the detailed 
analyses in the following chapters (I refer especially to the analysis of 
heads and premises in 13. 11) that it is the factual connection between 
referents, as apprehended by speakers, that is decisive. From that 
point of view there is no "leap" from 'prayers' to 'little balls', 
without intermediate stages, but on the contrary a very intimate asso- 

^) This wotdd involve the belief that adult people never add any new 
meaning to the range of the word that they have learnt — an obviously untenable 

'') Understand the phrase as a whole — that is just what all hearers do, adult 
or not. 


ciation between the two notions. And the intermediate stage is there 
too, in the form of a certain type of phrase, as obligingly pointed out by 
Jespersen himself. There is, then, no necessity of having recourse to 
misapprehensions, whether by children or by adults. With regard to 
the former, I may add that the sense-change in premises is exactly 
of the same type as that in heads: are we to assume that a legal term 
like premises is influenced by childish misunderstandings? And if 
not premises, why beads? 

Similar observations apply to Jespersen's assumptions (Lang. 172 — 
173) concerning phonetic associative interference (popular etymology), 
which he attributes to "childish confusion of unrelated words". We 
find that in the instances that can be verified in the NED, the sup- 
planted form is obsolete or obsolescent, at least in the relevant meaning; 
that is to say, it is no longer supported by a strong linguistic tradition, 
but can easily work loose, and it is consequently liable to be associated 
with some better known group. This is not a childish peculiarity, but 
common to all ages (cf. 9.1 and 9.4). 

Jespersen also adduces the word jain: "There were no connecting 
links between the meanings of 'glad' and 'obliged, forced', but when 
jain came to be chiefly used in combinations like he was jain to leave 
the country, it was natural for the younger generation to interpret 
the whole phrase as implying necessity instead of gladness" (1. c. 
175). Let us take a modern instance: suppose a man gets a defective 
copy of a book, and returns to the bookshop in order to get a perfect 
copy, saying, "/ should he glad to have another copy" — how young 
must the bookseller be in order to understand that this really means 
'I insist on getting a good copy, because I have paid the full price'? 
Is it not highly probable that he would understand it perfectly, even 
if, or rather because, he has outgrown the childish ignorance of his 
mother tongue? Jespersen has failed to see that the speaker may intend 
a phrase such as that with jain to mean 'he was obliged', and to under- 
stand it like that is then not to misunderstand, but to understand 
correctly. Moreover, the connecting link is there, in the form of the 
expression that Jespersen quotes. It is true that there is no logical 
"middle term", but that is entirely irrelevant. 

The second assumption is especially represented by Meillet, the 
assumption that there is no cause of sense-change strong enough to 
account for the constant flux of meanings if the discontinuity of 

l82 GUSTAF STERN 7.42 

tradition did not add its effect. My standpoint is quite different. I 
assume that the essential cause of change is the fact that the purposes 
for which speech is used are constantly setting new tasks to speakers, 
compelling them to adapt the instrument — speech — to the new task in 
order to cope with it successfully. Sense-change is therefore not only 
normal, but necessary (cf. 7. 11), and the functions of speech are causa 
sufficiens. It is no doubt true that in not a few cases the younger 
generation employs words in another way than do their parents. But 
that is explicable as an instance of the deviations usual in group speech, 
and due to the special interests common to the group; moreover, it 
applies to a limited number of words only, referring to matters in 
some way connected with these interests. "The younger generation" 
is a term to be taken cum grano salis.^) It is mostly some portion 
of the younger generation that makes itself conspicuous by linguistic 
freaks, e. g., the smart set, or the sporting set, or the more or less 
criminal set, and so on. Many of their catchwords are ephemeral, a 
few only are more tenacious, and end by becoming standard. 

In every class of change there are numbers of instances belonging 
exclusively to adult speech, on which children cannot possibly have a 
predominant influence. Why should not the same normal development 
be assumed with regard to words which might conceivably have been 
influenced by children? Saoul is an ordinary case of the fading of a 
euphemism (cf. 14.56). It is obvious that any adult speaker using 
such a euphemism about a drunken individual will have the notion 
of 'drunk' in his mind, and thus will associate it to the word saoul. 

I conclude, then, that the writers criticized here represent an 
antiquated opinion concerning the nature and causes of sense-change. 
They ask for logical affinity between the primary and secondary 
meanings. They have failed to adopt the modern explanation of 
semantic changes as due to the functions of speech, and are therefore 
driven to seek for causes in the wrong direction. I do not deny 
that misunderstandings occur, but I believe that Paul is right in 
restricting them to words not supported by a strong tradition, and so 
easily affected by associations that are not historically justified 
(cf. 13.12, note). 

7.43. The Part of Emotions in Sense-change. Wundt and Falk regard 
emotional sense-changes as a separate class, "both psychologically 

1) Rien de plus obscur que cette notion de "generation". Delacroix 189. 


and historically independent and in several respects incommensurable" 
(Falk 56). It will be apparent from the treatment in the following 
chapters that I consider the emotional changes as belonging to the 
same types as cognitive changes. The points of view that I have 
selected as my basis of classification cover changes in all kinds of 
mental material. Emotive changes are especially numerous among 
nominations (the figures of speech), and among adequations, the latter 
being caused by shifts in the subjective apprehension. Among the 
transfers there are, by definition, no emotive changes. Otherwise 
they may occur in any class. 

A still greater importance is ascribed to the emotions by Sperber. 
His theory, as stated in his latest book {Einfiihrung) may be sum- 
marized as follows: In order to explain sense-change it is necessary 
to discover in language a source of energy, the action of which may 
bring about the development (1. c. 30). This energy must be con- 
siderable, for millions of speakers must resolve to use a word in a 
modified sense before the innovation becomes definitely established in 
use. All these individual decisions must tend in one direction (1. c. 
32). In a large number of instances what we call emotional tone 
(feeling-tone) plays a decisive part. "Als psychische Energiequellen 
sind aber, in Gegensatz zu den blossen Vorstellungen, Gefiihle jeder 
Art ohne Zweifel anzuerkennen" (1. c. 37). For language is not only 
a means of communication, it is in quite as high a degree a means of 
expressing emotion.^) The latter function is more original than the 
former. A word with a strong feeling-tone does not behave as mere 
passive material but has, as it were, an active existence, often pres- 
sing itself upon the speaker from within, even if there is no objective 
motive for using it (1. c. 39. Cf. Stern, lyitteris III 58 sqq.). 

Sperber has tried to find the main cause of sense-change in another 
direction than Jespersen and Meillet. He assumes that speech has 
two functions only, expression and communication, and he does not 
employ the modern theory that speech is an instrument for the pro- 
motion of purposes. It is no doubt true that the sources of psychic 
energy are emotive, but this does not prove that emotion must 
involve sense-change. There are, after all, in every language a very 
large number of words for expressing emotions, and the necessity 

1) It seems almost as if Sperber assumed that only cognitive elements are com- 
municated, and only emotions expressed — which is of course erroneous. 

184 GUSTAF STERN 7,43 

of forming new expressions is not inherent in tlie need for expression 
as such. Indirectly, the expressive function, as well as the other 
functions, is a cause of change in that it may require an adaptation 
of the instrument to the momentary purpose. 

The assumption that words have — even if only metaphorically 
speaking — an active existence, is a return to an antiquated terminology, 
and should be rejected. The unsatisfactory state of the psychology 
of emotions in general is of course felt in the analysis of emotive 
sense-change. No adequate analysis is possible until psychologists 
have done their part of the work. 

7.44. Conservative Factors. Falk (Betydn. 54 — 55) states that 
the strongest conservative influence with regard to meanings is the 
association of a word with its cognates, when these retain their 
original meanings (cf. Stocklein 8). 

Meillet (lying. 236 — 237) thinks that when a word is no longer felt 
as belonging to its etymological derivational group the way is open 
for its meaning to change indefinitely. Thus vif is no longer felt as 
belonging to vivre and vie, and therefore the meaning 'mobile, anime' 
may come to dominate. 

It seems to me that it woiild be difficult to prove that the connection 
with the cognate group was severed before the change set in. It is 
much more likely that the severance is the result of the sense-change. 
Sometimes a phonetic change may give the word a form diverging 
from the norm of the group, and it is possible that this fact may 
facilitate a sense-change. As far as I know it has not yet been proved 
that, ceteris paribus, formally isolated words are really subject to 
change to a greater extent than others, which retain the characteristic 
form of their group. Difference of form is no obstacle to group-forma- 
tion (cf. 9.1). And there are innumerable instances of words 
retaining their form, but separating semantically from their group. 
I shall only mention the typical group orare — oratio — orator, where 
the phonetic similarity has not prevented the verb from going its own 
way (cf. also Nyrop's criticism, IV 75). 

I am inclined to believe that the preservation of meanings as well 
as their change is due to the same general cause: the functions of 
speech. I have already pointed out that the communicative and 
symbolic functions exercise a conservative influence in maintaining 
the stability of meanings. In order to be comprehended as swiftly 


and accurately as possible it is necessary to keep within the traditional 
range of the words, that is to say, to use them in their established 
meanings. The strength of this factor is shown by the reaction of 
words on thought, as described in 5.53, and by the importance of 
words for the formation of general notions and categories. 

As a second conservative factor, I regard the frequency of words in 
a specific meaning; in other words, the strength of the linguistic tradi- 
tion. It is well known that the most common words of a language 
retain most tenaciously old and otherwise discarded forms and inflec- 
tions. It is reasonable to assume that a strong tradition has similar 
effects on meanings. Note, however, that the retention of one or more 
old meanings is no obstacle to the acquisition of new ones: frequency 
is only a conservative factor for already established meanings. 

7.45. The Regularity of Sense-change. Semantic Laws. It is not 
much use quoting the contradictory opinions of philologists concern- 
ing the regularity of sense-change and the possibilities of formulat- 
ing semantic laws, more or less corresponding to sound-laws. Most 
of these opinions are pessimistic. I shall give instead an instance of 
what may be accomplished in favourable circumstances. 

The only method of ascertaining definitely the existence of regu- 
larities in sense-change is to study empirically and in detail, suitable 
groups of synonyms, and to find out to what degree their development 
runs in parallel lines and is conditioned by identical factors. Such 
attempts have been made more than once, but with inadequate material. 
It has been found that there are parallel developments, but no exhaust- 
ive account of any group of synonyms has, as far as I know, been 
attempted. At the present moment, I have complete material for 
one group only, in English, and I have given an account of it in a 
paper some years ago. Since most readers will not be able to see the 
paper I shall here summarize the main points (Stern, Betydelselag. 
See also Stern, Swift 6 sqq., 24 sqq., 208 sqq., 262 sqq.). 

The change to be investigated is that from 'rapidly' to 'immediately' 
occurring in a number of English adverbs. The change may be illus- 
trated by the following quotations: 
I. He wrote quickly. 
II. When the king saw him, he quickly rode up to him. 

III. Quickly afterwards he carried it off. 

This is a typical permutation (ch. 13), where the change takes place 


in phrases permitting a double interpretation (equivocation, of. 13.12). 
In I, the verb is imperfective, and the adverb means 'rapidly' (swift- 
ness in space). In III, the verb is apprehended as punctual (perfective), 
and the adverb means 'immediately' (swiftness in time), no attention 
being paid to the circumstance that the act must necessarily take some 
time to perform. 

In II, the meaning of the verb may, according to circumstances 
and context, be apprehended as imperfective, denoting the progress 
of the action, or as perfective, denoting the action as a unit. In the 
former case, the adverb means 'rapidly', as in I. In the latter case, 
it means 'immediately', as in III. Phrases of type II occur especially 
when to an otherwise imperfective verb, is added a word stating the 
purpose or end of the action, so that it is limited in time. If a person 
rides rapidly up to another, the action is completed within a short 
space of time. The equivocal instances thus do not represent a separate 
shade of meaning, but the adverb may be interpreted either in one 
way or the other, either as I or as III. 

In some of the adverbs in question, which are well instanced in the 
texts, the development can be traced in detail. In other cases the 
three types appear almost simultaneously, or even in another order 
than that indicated above. It might then be assumed, either that the 
two meanings 'rapidly' and 'immediately' have arisen in completely 
different ways, and that the equivocal meaning is the result of a con- 
tamination between I and III; or that the development has proceeded 
from 'immediately' to 'rapidly'. 

The former case, independent development of the two meanings, 
sometimes occurs, as for instance in fast (cf. Stern, Swift 93 — 94), 
which is therefore left aside in the following discussion, but as far as 
I have been able to see, not in any other of the adverbs mentioned 
here. It is of course necessary to investigate the whole history of 
each word in order to ascertain how the sense 'immediately' has arisen. 

With regard to the other possibility, a development from 'immed- 
iately' to 'rapidly', I have shown in Swift (218 — 219) that it is not 
probable, for various reasons. Sandegren (106) has arrived at the 
same result for German. As mentioned, it is evident that if a person 
rides rapidly up to another, the action is soon completed; but we 
cannot reverse the argument and say that if a person soon rides up to 
another, then the action is also rapidly performed. There is not a 


single instance in English of a development 'immediately' > 'rapidly', 
and on the whole the meaning 'immediately' appears to have rather 
limited possibilities of development. I know of only two developments 
from 'immediately', not counting the change from 'immediately' to 
'soon', which should be explained as a loss of intensity. One is the 
peculiar development of anon, in reply to a call, 'presently, coming 
at once', to 'at your service' > 'what did you say', a development 
which is obviously conditioned by a peculiar context in which the phrase 
is used, not by the meaning of the word in itself. The other instance 
is the development of the sense 'early' in soon. It has probably taken 
place in the comparative sooner, and afterwards been analogically 
transferred to the positive. Cf. a similar development in rather (Stern, 
Swift 34 sqq. and 9.24 below). 

We may therefore, I think, safely assume that when the two 
meanings 'rapidly' and 'immediately' occur in an English adverb, the 
former of them is the origin of the latter, in the manner described for 

The reason why the two senses are found in another chronological 
order than the one I have indicated, is especially the scantiness of 
the OK and ME texts. In OE some dialects are not at all represented 
by texts, and others only fragmentarily; the same is the case for early 
ME. It is only about 1300 that there is a satisfactory supply of texts 
to illustrate the state of the language. The fact is reflected in the 
frequent occurrence of the date 1300 in the lists below. In other 
cases, a meaning may have arisen in colloquial language, which is 
scarcely represented in our texts. Also, in some cases, the words are 
comparatively rare. In these circumstances, it is evident that a 
meaning may be much older than the earliest preserved record, and 
the latter should therefore be regarded as a terminus ad quern. We 

1) The NUD distinguishes the meanings of quickly in another way. Sense II 
is not given separately, which is only natural, since equivocal meanings are 
avoided by lexicographers. Sense III is divided into two, (i) denoting that the 
whole action or process is begun and ended within a comparatively short space 
of time, and (2) denoting that there is little or no interval between a given point 
in time and the doing of an act or the happening of an event. Logically, the two 
senses are easy to distinguish, but in actual fact there are very few cases which 
cannot be interpreted either way. Nor can they be distinguished chrono- 
logically and genetically. I have therefore not separated them here. Cf. 
Stern, Swift 25. 


have to be satisfied if the chronolog}^ of the well evidenced words caa 
be traced with certainty, and if the less frequent words do not present 
divergences that contradict the assumptions made. 

The adverbs in question show the following earliest known instan- 
ces.^) The OE period is taken as a unit, since the literature is too 
scanty for the circumstance that a word is found only in late OE to 
have much importance. 

Sense I Sense II Sense III 

'Rapidly' 'Rapidly/immediately' 'Immediately' 

Hrcsdlice OE OE OE 

Hrape {Rape) OE OE OE 

Ardlice OE OE OE 

Lungre OE OE OE 

Ofstlice OE OE OE 

Sneome OE OE OE 

Swipe OE OE 1175 

Swiftly OE OE 1200 

Caflice OE OE 1370 

Swift OE 1360 1300? 1400? 

Georne OE 1290 1300 

Hijendliche 1200 1200 1200 

Quickly 1200 1200 1200 

Smartly 1290 1300 1300 

Snelle 1300 1275 1300 

Quick 1300 1290 1300 

Belife^) 1200 1200 1200 

Nimbly 1430 1470 1400 

Rapely 1225 1300 1325 

Skete 1300 1300 1200 

Tite 1300 1350 1300 

Wight 1300 1360 14th cent. 

Wightly 1350 1350 1300 

^) I cannot reprint here all the material on which my dates and conclusions are 
founded; I have to refer the reader to Stern, Swift 209 sqq., and Stern, Betydelselag. 

2) With regard to belife, it is probably due to chance that no earlier instance 
of sense I is known. There are instances in I,ayamon A and Ormulum which 
appear to represent the intermediate stage between 'vigorously, eagerly' and 
'rapidly'. The existence of the latter sense at least about 1200 may therefore 
be safely assumed. 


I have no certain cases of the development 'rapidly' > 'immediately' 
except in the words enumerated above. We may therefore state that 
this development has, in all cases, taken place before 1400. 

We have next to give a list of the adverbs meaning 'rapidly' but 
not 'immediately'. I leave out words in which the sense of speed is 
mixed with other elements of meaning which may have influenced the 
■development. See the lists in Stern, Swift 6, 8, and 209. The dates, as 
before, give the earliest known instances. 

speedily 1300 Expeditiously 1603 

Speedly 1300 Postingly 1636 

Speedfully 1398 Speedingly 1647 

Rashly 1547 Velociously 1680 

Roundly 1548 Rapidly 1727 

Post 1549 Postwise 1734 

Amain 1563 Hurryingly 1748 

Post-haste 1593 Hurriedly 1816 

Fleetly 1598 Fleetingly 1883 

We find that, with three exceptions, adverbs meaning 'rapidly' but 
not 'immediately' have acquired the former sense only after 1400. 

If such a chronological distribution of instances had been ascertained 
with regard to a phonetic change, we should at once attempt to explain 
the over-lapping words as due to some collateral influence. 

The three adverbs are derived from the noun speed, OE. sped, 
originally signifying 'abundance, prosperity, success'. The meaning of a 
derivation is naturally influenced by that of the stem word, as long 
as they are felt to belong together (cf. 9.25). The adverbs may have 
preserved elements of their original meaning, which prevented a di- 
verging development. Another explanation might be that of insuffi- 
cient frequency. The change could be effected in certain contexts 
only, and on condition that these contexts occurred with sufficient 
frequency. That was perhaps not the case here. 

Note further that the words which are found with the sense 'immedi- 

I repeat that the dates are to be taken as termini ante quos. Some of the 
words are comparatively rare. I have only nine instances in all of swift, and 
only eight of nimbly. In such circumstances it may be a mere chance that the 
various meanings are not found earlier. 

With regard to loan-words we have to reckon with the possibility of the sense 
'immediately' arising in the foreign language, not in England. This might be 
possible for tite. 


ately' only in the 14th century, are all comparatively rare. It is, I 
think, not improbable that all developments 'rapidly' > 'immediately' 
happened before 1300. The date 1300 for speedly and speedily is of 
course also a terminus ante quern, but it is perhaps not altogether 
impossible that these words did not acquire the sense 'rapidly' until 
the tendency to change had disappeared, and consequently fall outside 
the time limit. But this explanation is more uncertain. 

We have found that English adverbs with the sense 'rapidly' are 
divided into two chronological groups, one in which the sense is earlier 
than 1300 (or 1400) and in which the sense 'immediately' nearly always 
arises out of it; another in which the sense 'rapidly' is later than the 
date mentioned, where no such development occurs. It is further 
demonstrable that the development always takes place in definite con- 
texts: when the adverb is employed to qualify verbs which may be 
apprehended as imperfective or as perfective (punctual). We may 
therefore formulate the following semantic law: 

English adverbs which have acquired the sense 'rapidly' before 1300, 
always develop the sense 'immediately'. This happens when the adverb 
is used to qualify a verb, the action of which may be apprehended as 
either imperfective or perfective, and when the meaning of the adverb 
consequently is equivocal: 'rapidly/immediately'. Exceptions are due 
to the influence of special factors. 

But when the sense 'rapidly' is acquired later than 1300, no such 
development takes place. There is no exception to this rule. 

This "law" has the form of a sound-law: it gives the circumstances 
of the change and a chronological limit. 

We ask, next, what may be the reasons for the cessation of the de- 
velopment. It cannot have been that the conditions favouring it ceased 
to exist, for we may still say, he went rapidly out of the room; but this 
has not caused a change of meaning for rapidly. 

It seems that the tendency itself disappeared. The reason is obscure. 
The changes began at a period when OE, without any consider- 
able influence from other languages, was following its own line of 
development; they continued during the periods of Scandinavian and 
French influence, and ceased as the importation of French and Latin 
linguistic material was at its height. We cannot demonstrate any 
connection between the general linguistic and cultural development, 
and the sense-change in question. 


We might perhaps point to the fact that the demand for means of 
expressing the notion of 'immediately' must have been satisfied by all 
these words. It is true that some of them soon disappeared but there 
were many others denoting the same thing, and the notion 'immedi- 
ately' has not so many different characteristics that it would require a 
great number of synonyms to be adequately represented. The tendency 
can therefore cease without detriment. We find that the tendency was 
effective as long as there was material for it to work upon. About 1300 
the change had been accomplished in all the relevant adverbs then 
existing. The s/)^^^- derivations were next in time, making their appear- 
ance probably during the 13th century. Perhaps the tendency then 
had weakened or completely disappeared, but even if that was not the 
case, these words might still have escaped it owing to their low fre- 
quency, or because their meaning w^as blended with other elements. 
After this, a considerable time elapsed before any fresh word acquired the 
sense 'rapidly', and the tendency could then have completely disappeared. 

As the changes began before our earliest literary records, we know 
still less of their first appearance. 

When a number of these adverbs had acquired both meanings, ana- 
logical influence on the remaining adverbs might be assumed (see on this 
point 9.34). But there are several arguments against the assump- 
tion. Analogical influence ought to work with equal strength in both 
directions, so that words meaning 'immediately' receive the sense 
'rapidly', but there is no trace of such a development. And why has 
not the analogical influence continued during the Modern English 
period, with regard to the adverbs in my second list? 

It does not seem improbable that an investigation of a large number 
of sense-changes falling within one sphere of experience would show 
that tendencies like the one just described may affect several groups of 
synonyms. It ought then to be possible to ascertain with greater pre- 
cision both the extent of the tendency and its chronological limits, and 
also to ascertain if it has any connection with general cultural factors. 
At any rate, v/e should gain an important contribution to the history of 
the language. But laws like these cannot be formulated for nomina- 
tions, or substitutions, and probably not for analogies and shortenings. 
Moreover, I think it is rare for any large group of synonyms to be so 
untouched by substitutions as are the swift-words. It would probably 
be more diffictilt to formulate similar "laws" for other groups. 


8.1. Theoretical Discussion. Substitutions are sense-changes due 
to external, non-linguistic causes.^) 

The process of change may be analysed as follows. Horns, real 
animals' horns, were used for the purpose of producing certain sounds 
and signals. When they were so used, the main thing about them was 
their function; and the fact that they were manufactured from animals' 
horns was not attended to. The notion of 'musical instrument' thus 
became the predominant element of meaning. This is a process of 
adequation, as analysed in 14. ii (cf. also the diagram given there). 
When people began to manufacture, from other material than horn, 
instruments of similar shape, and capable of producing similar sounds, 
in other words, instruments with a certain function, these were, as a 
matter of course, apprehended as belonging to the category of horns, 
and were consequently so named. This is the substitution. It is fol- 
lowed by a fresh adequation (cf. 14.14). 

The main condition for substitutions thus lies in the incessant modi- 
fication to which all objects, qualities, and actions are exposed, which 
are in any way connected with the momentary status of civilization, 
material, intellectual, or moral. A second condition is our method of 
arranging all items of experience into categories, each of which generally 
has a name of its own. 

The causes of substitution lie in the fact that referents change and 
that we require new names for them; these we get, in the present case. 

^) The term substitution is a part translation of Wellander's term Bedeutungs- 
unterschiebung (Wellander, Studien I 55 sqq., 70 sqq.) The name is justified by- 
regarding the change as caused by the substitution of a new referent. Collinson, 
MLR 20, 1925, 102, suggests the term subreption. Wundt discusses these shifts 
under the heading Historische Interpretation (II 477 sqq.). See further Nyrop 
IV 84 sqq., Marty, Unt. 543 sqq., Stocklein 34 sqq., Hatzfeld, Leitfaden 8i sqq., 
and the other works quoted by these writers. 


by placing the referent in some known category, denoting it by the 
same name. 

Substitutions are sometimes difficult to distinguish from regular 
transfers (ch. 12). The essential difference may be illustrated by com- 
paring the instance above, horn 'musical instrument', with a transfer, 
as for instance saddle 'rider's seat' applied to a mountain ridge of a 
certain — saddle-like — shape. In the former case, the metal or ivory 
horns — the new referents — and the old horn '^'of horn" form one 
category, and are therefore naturally given the same name. In the 
latter case, saddle 'rider's seat', and saddle 'mountain ridge' of a certain 
shape, do not form a category. The distinction of substitutions and 
adequations is discussed in 14.14. I refer, also, to the alternation of 
substitutions and adequations described in 14.11. 

Substitution is an extremely frequent form of semantic change. There 
are few groups of synonymous or cognate words which have not been more 
or less affected by it. The stock of meanings in a language reflects in a 
thousand ways the momentary state of the material, intellectual and 
moral civilization of the speaking community. The constant progress 
and modification of all forms of human life and thought re-act on the 
meanings. 'In the course of time, such modifications of meaning amount 
to considerable sense-changes, even if the change is gradual and at any 
one moment hardly perceptible. 

Although substitutions are very frequent there appears to be some 
words that are seldom touched by them. Wellander (Studien 1 100) 
assumes that substitutions occur mostly with concrete referents, but 
this is no doubt a mistake. The important point is not the degree of 
concretion or abstraction, but whether the referent is in any way, directly 
or indirectly, exposed to the action of cultural evolution. Concrete 
referents like parts of the body, or the actions denoted by verbs like 
run, go, eat, or by adjectives like hard, soft, big, little, etc., have remained 
practically untouched by substitution. The same is the case with 
notions like light, darkness, day, night, sleep, dream, and many others. 

For many philologists substitutions are the most interesting kind of 
sense-change because they help to throw light on the facts of human 
history. The development of meaning has followed the development of 
the referent, and if we can trace the former we can perhaps, at least in 
some cases, draw conclusions with regard to the latter, and vice versa. 
Worter und Sachen, ohne Sachjorschung keine Wortforschung, are well- 

Goteb. Hogsk. Arsskr. XXXVIII : i 


known formulae. The matter is familiar to every philologist, and since 
it falls outside the scope of my work, I may leave it with this remark. 

From the linguistic and psychological point of view which I apply, 
the substitutions are not much differentiated, and they are of no very 
great interest for my purposes. I shall content myself with a brief 
discussion, referring for further details to the writers quoted above. 

We may, I think, distinguish three main types of substitution, accord- 
ing to the origin of the change. The origin may be a factual change of 
the referent, or a change in our knowledge of the referent, or a change 
in our emotive attitude towards the referent. 

8.2. Factual Change of the Referent. In this type of substitution 
the referent is subjected to a factual change, due to progress or modifica- 
tions of technique, habits, etc. The referent may be concrete or ab- 
stract, and the change may be gradual or sudden. 

When a steam-engine was first installed on board a ship to function 
as an auxiliary propelling power, the ship itself was not otherwise 
changed: it looked as usual, with the exception of a narrow funnel. As 
the machine gradually superseded the sails, the appearance and con- 
struction of the ship slowly changed. At present a sailing ship and a 
steam-ship are generally quite different in appearance, but that is the 
result of a gradual development, during which there has never been any 
reason for not placing the new referent in the category of ships. 

Similar slow changes, due to technical progress, have affected an 
enormous number of objects, as house, carriage, machine, motor-car, 
tramway, telephone, telegraph, war, trade, and so on. The corresponding 
adjectives and verbs have been similarly affected. The word artillery 
originally signified 'warlike munitions, implements of war; ammunition 
in the wide sense', and more especially 'engines for discharging missiles, 
including catapults, slings, arbalests, bows, etc.': Jonathan gaue his 
artillery vnto his ladde, and said vnto him, Goe, cary them to the citie (i 
Sam. XX 40. 1611). The invention of gunpowder led to the construction 
of engines of war of a new kind; nevertheless, the junction being the 
same, the new machines were apprehended as belonging to the same 
category, and they were consequently denoted by the old name (cf. 
ISTED, Hatzfeld, Leitfaden 83, and Nyrop IV 84). The booking-office 
has its name from the practice of booking one's name for a coach or 
other conveyance. After the introduction of railways and steamboats,, 
the system of tickets was adopted. The place where they were sold 


was also called a booking-office, because at a booking-office one paid 
for a journey in some conveyance, and that is the main thing from 
the travellers' point of view. Mews is the plural of mew 'a cage for 
hawks'. The Royal Mews were at Charing Cross in London; the name 
was apprehended as attached to the place and was retained for the 
Royal stables, which were later built on the same site (earliest 
quotation from 1529). After adequation the name was transferred to 
other similar establishments, and the present meaning is 'a set of stab- 
ling grouped round an open space, yard, or alley, and serving for the 
accommodation of carriage-horses and carriages' (earliest instance in 
the NED from 1631). 

In the instances artillery and booking-office, as in that of horn, a 
process of adequation is a necessary preliminary to the substitution. 
A booking-office is, as the name says, a place where one books a 
passage with a conveyance; if that element of meaning had remained 
predominant, or at least prominent, the name could obviously not 
have been applied to a place where tickets are sold, but no booking 
of names done. The term must first, through adequation, have come 
to be apprehended as denoting a place where one pays for a passage. 
Similarly, artillery must have been apprehended as meaning, primarily, 
'engines of war used to throw missiles' (not necessarily by mechanical 
power) before the name could be used for guns charged with powder. 

It seems likely that we should assume a similar alternation of 
adequation and substitution in many other instances that are quoted 
in the textbooks; e. g. in the development of F. plume, G. Feder 
'quiir > 'pen' (Stocklein 34); the historical facts with regard to this 
word, are, however, not quite clear. 

In other cases, no preliminary adequation is necessary, and the 
change is sudden. Deutschland is not the same thing after 1871, as 
before that year; the treaty of Versailles and the republican constitu- 
tion have changed it again (cf. Wellander, Studien I 55).^) 

1) Wellander (Studien I 148) regards it as a case of Namengebung (intentional 
naming) when the German authorities fixed the hours between which night and 
day are to be reckoned for postal and other administrative purposes. But this 
is a clear case of substitution, since it is the referent that is arbitrarily regulated, 
while the name remains. On the other hand, when a totally new referent is cre- 
ated, as in the case of Mark (1. c. 151), and then named with an old word, that 
is a case of intentional naming. 


Here belong instances where abstract notions are defined anew, 
more or less differently, but the name is retained. A case in point is 
the definition of meaning in this book. (Cf. K. O. Hrdmann 34). 

The word cook (late lyatin cocus < coquus) 'one whose occupation 
is the preparation of food for the table' originally was mascu- 
line only. It is recorded in English from A. D. 1000. Applied to a 
woman it is not found till 1535 (NED). We may perhaps assume 
that this is a substitution, conditioned by a previous adequation of 
the word, in which the notion of the masculine gender receded or 
totally disappeared. 

A hoy is, according to the NED, "a male child below the age of puberty. 
But commonly applied to all lads still at school, as such; and parents 
or sisters often continue to speak of their grown-up sons or brothers 
as 'the boys' ". Is this international practice a case of substitution? 
Girl is similarly used. 

In this connection I may mention the change of meaning that 
occurs in christian names in the lapse of time. When Charles Smith, 
the baby, is baptized, his name has not the same meaning as fifty 
years later, when he may be an admiral, or a well-known author, or an 
important business man. The change is due to the change of the re- 
ferent and his various activities, and must consequently be classed here. 

The modification of habits is reflected in the change of meaning of 
the names of the daily meals in the Romance languages, as shown in 
Herzog's dissertation on the subject. In the Middle Ages, in France, 
there were two chief meals, disner (more rarely called desjeuner), about 
9 a. m., and souper, rarely cene, late in the afternoon. The disner 
was gradually taken later, and an early meal was called desjeuner. 
During the reign of Francis I, diner was eaten between 9 and 10, 
during that of Henry IV between 11 and 12, and in the 17th century 
about noon. Madame de Sevigne writes in 1676: elle aimerait bien a 
vivre reglement et a diner a midi comme les autres. In 1782 a writer 
states: a trois heures on voit pen de monde dans les rites, parce que 
chacun dine. At present, dinner is often eaten at eight or nine p. m., 
while the supper may be a light meal after the theatre, and the 
dejeuner has been changed into a midday meal. Herzog adds that 
this development is often limited to the cities, while in country 
districts an earlier arrangement is still preserved. The NED gives the 
following definition of dinner: "The chief meal of the day, eaten 


originally, and still by the majority of people, about the middle of the 
day (cf. Germ. Mittagessen), but now, by the professional and fashionable 
classes, usually in the evening." 

The gradual modification of political conditions is reflected in the 
different notions attached to the word king, if we compare, for instance, 
the period of William the Conqueror with that of George V. 

The meanings of ethical, aesthetic, religious, philosophical and 
other scientific terms are in a constant flux (cf. Paul, Prinz. 105). I 
I shall only quote the words religion, God, sacrifice, holy, and so on, 
which, together with their equivalents in other languages, have had 
their meanings greatly modified by the introduction of Christianity. 
Some of these changes, or at least some phases of the changes, should 
perhaps be placed in 8.4; cf. Schreuder 126. 

8.3. Change in Knowledge of the Befeient. The second type of 
substitutions includes instances where the referent in reality remains 
unchanged, but our knowledge of it changes. 

The progress of scientific knowledge has led to new notions being 
attached to many referents, and consequently to new meanings for 
many words, as electricity, solar system, atom, the South Pole, the Stone 
Age, Ancient Mesopotamia, to mention only a few. 

An instance of a somewhat different type is the word creek. It is 
originally 'a narrow recess or inlet in the coast-line of the sea, or the 
tidal estuary of a river; an armlet of the sea which runs inland in a 
comparatively narrow channel and offers facilities for harbouring and 
unloading smaller ships;' and, as an occasional extension of this sense, 
'an inlet or short arm of a river, such as runs up into the widened mouth 
of a ditch or small stream, or fills any short ravine or cutting that 
joins the river'. In the U. S. and British Columbia we find in 
addition the following sense, 'a branch of a main river, a tributary 
river; a rivulet, brook, small stream, or run'. The NKD assumes that 
"probably the name was originally given by the explorers of a river 
to the various inlets or arms observed to run out of it, and of which 
only the mouths were seen in passing; when at a later period these 
'creeks' were explored, they were often found to be tributaries of great 
length; but they retained the designation originally given, and 'creek' 
received an application entirely unknown in Great Britain". If this 
explanation is correct, we have here an instance of substitution due 
to increased knowledge of the referent. 


8.4. Change of Attitude to the Referent. We have, finally, the 
substitutions that are due to a change of attitude to the referent, a 
change of attitude that, in its turn, is due to non-linguistic, external 

A case in point are political catchwords. White Rose or Red? Home 
Rule, Corn Laws, Repeal, Women's Suffrage, have all, in their time, 
been fraught with emotion for many speakers, but at present they are 
probably less explosive. "There was a time during the Indian Mutiny 
when the name of Tantia Topi would kindle any Englishman into a 
blind rage. But who cares for Tantia Topi now? " (Newspaper). 

When a thing falls into disrepute owing to a change in the general 
attitude towards it, the meanings of the relevant words suffer accord- 
ingly. This happened with the philosophical system of the Middle 
Ages, scholasticism, and we find the word signifying: 'servile adherence 
to the methods and teaching of the schools; narrow or unenlightened 
insistence on traditional doctrines and forms of exposition': He 
found his country tied up in formalism, scholasticism, and tradition, 
and by strokes as remarkable for boldness as strength he set it free (1861). 
The adjectives scholastic and scholastical have had to submit to the 
same fate: Sidney's Arcadia is not romantic but scholastic, not poetry, 
but casuistry (1820). A proper distinction, by the ivhiche you may 
escape the scholasticall snares and mases (1531). Cf. the remarks on 
dunce (14.84). 



9.1. Group Formation. 

9.11. Preliminary Remarks. Analogy is the traditional name for 
a very frequent linguistic phenomenon, in Germany also called System- 
zwang (systematizing or levelling tendency). Philologists writing 
on analogy^) have mostly concentrated on the morphological and 
phonological aspects of the question, leaving semantic problems aside. 
To some degree they are inseparable, but I shall as far as possible 
limit my discussion to the analogical change of meanings. 

The linguistic material for the present chapter is very unsatisfactory. 
Analogy is often assumed, but the assumption is seldom really proved. 
In some cases it is of course possible to draw a more or less certain 
conclusion on the basis of a comparatively superficial knowledge, but 
in other cases the only real proof would be a historical study of the 
word in question, showing that the relevant meaning cannot be ex- 
plained as originating from any other known meaning of the same 
word, but must be due to outside influence. There is very little of 
this kind of material available. 

Analogy is primarily unintentional, but any one of the formations 
described here may also be produced intentionally. 

Analogy is defined by Hoffding (Analogi 34) as "identity of rela- 
tions between separate objects, not identity of the single character- 
istics". We shall find that the effect of the two first groups of 
analogical change described below, combinative analogy and cor- 
relative analogy, is to bring about such an identity of relations within 
semantic groups. With regard to the third group, phonetic associative 
interference, it can be classed as analogy only if we take the definition 

1) Thumb and Marbe, Exp. Unt. i sqq. give a useful summary of the literature 
on analogy; see also B. I. Wheeler, Analog3% Misteli, Zs. f. Volkerpsych. 11 (1880) 
443, Paul, Prinz. 35, 106 sqq., 189 sqq; R. M. Meyer, Zs. f. vgl. Sprachf. 43, 
352 sqq., Pos 122 sqq. 


in a wide sense. Associative interference leads to the formation of 
semantic groups, and in so far as speakers instinctively try to get 
isolated words into one group or another, that is also a kind of analogy. 

The problem of the semantic groups just mentioned is of essential 
importance for the matter in hand, and these groups have generally 
been regarded as the cause of analogical change. It will be useful to 
begin the discussion with an investigation into the nature and extent 
of group formation, and the part played by it in language. The current 
opinion is expressed most precisely by Paul, and I shall first give a 
short summary of his views. Wundt's explanations, although for- 
mulated in different terms, are not essentially different from Paul's 
(cf. Wundt I 443 sqq.). 

9.12. Paul's Theory of Analogy. Paul (Prinz. 106 sqq.) assumes 
that words attract each other in our mind, thus forming larger or 
smaller groups of various kinds. Words of identical or similar basic 
meaning form stoffliche Gruppen, material groups. For instance, the 
case forms of a noun constitute a material group. Within this group 
the cases of the singular as contrasted to those of the plural constitute 
a smaller material group; and so on. Words of cognate meaning form 
a larger material group with laxer internal cohesion. The similarity 
of meaning is generally accompanied by similarity of form, usually 
conditioned by etymological affinity, as fuhren-Fuhrer-Fuhriing. 
There are also material groups composed of totally unrelated words: 
boy-girl, old-new, and so on.^) 

Formale Gruppen, formal groups, are constituted by words with 
similar or identical relational meaning, but varying basic meaning. 
Thus all nomina actionis, all comparatives, all nominatives, all first 
persons of verbs, and so on, constitute formal groups. 

Not only single words cohere in this way into groups, but also 
analogous proportions between words. The basis for the formation of 
Proportionengrnppen, proportional groups, is the crossing of material 
and formal groups, exemplified by series like Tag — Tages — Tage, 
Arm — Armes — Arme; Fiihren — Fiihrer — Fiihrung, Er Ziehen — Er- 
zieher — Erziehung, and so on. 

^) Osthoff, Suppletivwesen 3 sqq., distinguishes unechie stoffliche Gruppen 
constituted by forms of different stems as bonus — melior — optimus, fero — 
tuli — latum, and echte stoffliche Gruppen, in which all the forms are of one stem. 
Cf. Bally I 39 sqq. 


The action of these groups "besteht dabei gewissermassen in der 
Auflosung einer Proportionengleichung, indem nach dem Muster von 
schon gelaufig gewordenen analogen Proportionen zu einem gleichfalls 
gelaufigen Worte ein zweites Proportionenglied frei geschaffen wird. 
Diesen Vorgang nennen wir Analogiebildung" (1. c. no). Paul in- 
sists on the point that an analogical formation is equivalent to the 
solution of a proportional equation, so that at least three members 
must be known. "Es muss jedes mit dem andern irgendwie vergleichbar 
sein, d. h. in diesem Falle, es muss mit dem einen im stofflichen, mit 
dem andern im formalen Elemente eine Uebereinstimmung zeigen. So 
lasst sich z. B. im L,at. eine Gleichung ansetzen animus: animi = 
senatus: x, aber nicht animus: animi = mensa: x. Es kann daher ein 
Wort in seiner Flexion von anderen nur dann analogische Beeinflussung 
erfahren, wenn es mit diesen in der Bildung einer oder mehrerer For- 
men iibereinstimmt. Es kommt allerdings zuweilen eine Beeinflussung 
ohne solche Uebereinstimmung vor, die man dann aber nicht mit 
Recht als Analogiebildung bezeichnet. Es kann eine Flexionsendung 
wegen ihrer besonderen Haufigkeit als die eigentliche Normalendung 
fiir eine Flexionsform empfunden werden. Dann iibertragt sie sich 
wohl auf andere Worter auch ohne die Unterstiitzung gleichgebildeter 
Worter" (Prinz. ii6 — 117). 

Paul's mathematical explanation is decisively, and I think rightly, 
rejected by Delacroix: "il faut distinguer I'analogie du raisonnement 
qui conclut en vertu d'une ressemblance entre les objets sur lesquels 
on raisonne. D'analogie linguistique ne consiste aucunement a calculer 
la quatrieme proportionnelle, a determiner un terme par la connais- 
sance de I'un des couples et d'un des termes du second; pas plus qu'a 
s'elever par I'observation des rapports a la raison des choses, ou a 
Her entre elles des ressemblances exterieures dont on ne connait point 
la raison. II n'y a pas de raisonnement dans I'analogie linguistique; 
tout au plus Taction de I'esprit qui continue spontanement un mouve- 
ment anterieur" (Delacroix 250. Cf. also Freudenthal 99 sqq.") 

I believe that Paul is on the right track in speaking of "die Vor- 
stellung einer Allgemeingiiltigkeit der Muster, welche dem Einzelnen 
das Gefiihl der Berechtigung zu eigenen Zusammenfiigungen gibt" 
(1. c. Ill), and I hope to show that the process mentioned in the last 
lines of the quotation from Prinz. 117, which Paul refuses to regard 
as a real analogy, is of essential importance. 

202 GUSTAF STERN 9. 1 2 

It is a weak point in Paul's theory that he has adduced no indepen- 
dent evidence in support of his assumptions concerning group forma- 
tion. The first thing we have to do is to find such evidence, and to 
see if group formation realh- extends as far as Paul asserts. (On the 
theory of group formation, see Cassirer, Phil. I 261 sqq.) 

9.13. Basic {Material) Groups. As the material groups are 
groups of words with identical or correlated basic meaning, I shall 
employ the term basic groups. Paul's formal groups will be called 
relational groups, since they consist of words with identical relational 

A flexional group is a basic group constituted by the various flexional 
forms of one word. When discussing the semantic problems we have to 
remember that each flexional form is generally capable of expressing 
more than one relation — for instance, a geriitive form may express 
possessive, objective, or subjective genitive — and that consequently 
the semantic group, the collocation of a basic meaning plus all the rela- 
tional meanings with which it may normally be combined, is much 
larger than the corresponding morphological group. The six tense 
forms of a L,atin noun are capable of expressing a much larger number 
of relations. 

The actual existence of flexional groups may be taken for granted. 
The cohesion within a flexional group is generally fairly strong, and it 
is exceptional for one member of it to separate entirely from the group 
with regard to semantic development. 

A derivational group is a basic group consisting of all the derivations 
of one stem that are still apprehended by linguistic feeling as belonging 

The coherence among derivations of one stem is not so strong as that 
between the flexional forms of one word (Paul, Prinz. 195, 205). One 
word may pass through a phonetic or semantic development which 
separates it from the cognate words. Thus, no Englishman without 
etymological knowledge apprehends to singe as a derivation of to sing. 
Different and indifferent have to some extent drifted apart in meaning, 
but the phonetic similarity will perhaps keep them together; similarly 
invasion and evasion. (Cf. Bally I 33, 39, 143 II 21, 22, Carnoy 139, 
Hatzfeld, Bedeutungsverschiebung 34, Stocklein 8). 

^) Groups are taken to mean only groups that are apprehended as such by 
linguistic feeling. 


The existence of derivational groups is also certain. We must only- 
take care to remember that single members of an etymological group 
may separate from it, so that etymological affinity is not sufficient to 
establish an actual affinity, perceptible to linguistic feeling. 

A correlative group is a basic group consisting of words the meanings 
of which are in some way correlated, and therefore apprehended by 
linguistic feeling as belonging together. Such groups are hoy-girl, new- 
old, go-come, up-down, summer -winter , and so on. (Cf. Wundt I 448). 

Linguistic evidence for the existence of such groups is found in the 
well-known fact that words are not seldom influenced in form, gender, 
etc., by other words of opposite or otherwise correlated meaning. Thus, 
OE mycel 'large' probably has its vowel from Ifftel 'small', the original 
stem vowel being -i-. Female has been remodelled from F. femelle un- 
der the influence of male. Late Latin grevis for gravis is due to levis; F. 
rendre from L. reddere, to prendre; the masculine gender of F. ete 'sum- 
mer', from L- aestatem, fem., is due to the masculine gender of hiver, 
printemps, automne. Senexter is influenced by dexter, OE. fcfen by mor- 
gen, Germ. Morgend by Abend, and Olcel. fleire by meire. Further 
instances may be found in most handbooks (for instance Brugmann- 
Delbrtick II 1.17, Wundt I 447, Carnoy 55). 

The problem of the correlative groups has been investigated experi- 
mentally by psychologists. The experimenter calls out a word at a 
time, and the observer is instructed to respond with the first word that 
occurs to him. We may assume that, if there is any closer connection 
between two words, and one of them is used as stimulus, then the other 
will appear as response in a majority of cases. 

Thumb and Marbe (Exp. Unt. 17 sqq.) found that some words have 
favoured responses (bevorzugte Reaktionen) , occurring in more than half 
of all the instances. Esper, experimenting with English words, arrived 
at similar results (Psych. Rev. 25) 

Two words often function as favoured responses to each other. That 
is the case with adjectives of opposite meanings, as large-small, heavy - 
light, old-young, thick-thin, white-black (Thumb, IF 22, 22), or with 
complementary meanings, as brav-fleissig, einsam-ode, grau-griin, nord- 
lich-sUdlich (F. Schmidt 92); with pairs of names for personal relations, 
as father-mother , with pronouns, as I-you, with certain verbs, as give- 
take, and certain common adverbs, as here-there. Numerals are mostly 


associated with higher numerals; the numerals i — 10 mostly with the 
next higher numeral (Thumb 1. c). 

In some cases a third word appeared as ndchst bevorzugte Reaktion. 
Some words were absolutely favoured as responses, without regard to 
the stimulus word; these were words with a high general frequency in 
language (F. Schmidt 88 sqq., Eberschweiler 271, Dauber 180 sqq, 
quoting similar experiments by Saling and Reinhold, Zs. f. Psych. 49 
and 54). 

Psychological evidence thus corroborates the assumption that has 
been made on the basis of linguistic phenomena: that there are groups 
of two, three, or, in a few cases, more words, which are apprehended by 
linguistic feeling as belonging together. But there is not sufficient evi- 
dence to show whether such groups embrace a majority of the w^ords in 
a language, or if the majority are isolated. 

We have, then, basic groups of three kinds, (i) Flexional groups in 
which one basic meaning runs unchanged, or comparatively unchanged, 
through a number of combinations with varying relational meanings, 
forming together the flexional scheme of one word. (2) Derivational 
groups, in which, in a similar manner, one basic meaning runs through 
combinations with varying relational meanings (often expressed by 
suffixes or prefixes, ablaut or other modifications, sometimes not ex- 
pressed at all), forming together a group of derivations from one stem, 
a derivational scheme. And (3) correlative groups, consisting generally 
of two or three words whose basic meanings are correlated in such a 
manner that the words are often apprehended together. The names of 
the months and the numerals probably constitute the largest correlative 
groups. The names of the days in the week, and of the seasons, are 
other instances (Thumb and Marbe, Exp. Unt. 51 sqq.). It is not known 
what proportion of the vocabulary belongs to correlative groups. 

9.14. Relational {Formal) Groups. For the existence of relational 
groups there is no conclusive linguistic evidence. R. M. Meyer (Zs. f. 
vgl. Spr. 43, 352 sqq.) regards the re-appearance of a typical suffix, as 
Kluge has demonstrated for Germanic names of nations, mountains, 
houses, shops, bread, baskets, etc., as a sign that these words form a 
semantic system, and this is, apparently, a common opinion. 

I do not think that the conclusion is tenable. The fact that the nouns 
of a certain class, as for instance national names, have a common suffix, 
is the result of combinative analogy (see 9.22); B. I. Wheeler, Analogy 


I, thinks that all groups are the result of analogy, which is probably not 
correct. A small number, perhaps only one single combination — the 
number of originals is of small importance, according to Thumb and 
Marbe, Exp. Unt. 82 — may be the origin of the whole class. The suffix 
is apprehended as representing a certain relational or basic meaning, 
and is therefore combined with other stems in a similar way. At the 
moment of creation, each new name was possibly felt to be con- 
nected with the name or names, on the analogy of which it was formed. 
Suppose that the first compound is A; on the analogy of A were 
formed B, C, and D; on the analogy of B were formed E and F; on the 
analogy of C, G and H; and on the analogy of G, I and J; and so on. 
The meaning of the suffix would be liable to change during this process, 
and although the whole series, historically, forms one chain with many 
ramifications, the chain is not necessarily closed, that is to say, E, H, I, 
and J need not have any direct connection — to linguistic feeling — 
with A or with each other. Cf . as an instance of this, the history of the 
suffix -let in the NED. 

Experimental evidence shows that there is a tendency to respond with 
noun to noun, with adjective to adjective, and so on. The parts of 
speech are of course linguistic categories, and may be called relational 
categories, since each possesses specific syntactical functions, that is, 
specific relational meanings, but this fact does not prove that they form 
relational groups, i. e., that they are felt to belong to each other, 
as for instance the case forms of one noun. The tables in Thumb and 
Marbe show that the authors used stimulus-words having a favoured 
response (as young-old, etc., cf. 9.13), and this circumstance of course 
greatly increases the number of responses with the same part of speech. 
In other cases we should probably assume that the syntactical nature 
of the stimulus word turns the observer's attention in a certain direction 
and favours a response belonging to the same category (see Selz II 
432, Jung-Riklin IV 59 sqq., F. Schmidt 65 sqq., Esper, Ps. Rev., 
Dauber, Eberschweiler) . 

F. Schmidt (1. c.) experimented with verbs. His observers were 
schoolboys who had not yet been taught the conjugation, and who 
were therefore not influenced by the printed paradigms of grammars. 
Schmidt found that the first person singular present indicative was the 
most favoured response for nearly all other forms, when the observers 
responded with a form of the same verb as that used as stimulus. The 


result directly contradicts Paul's theory, according to which that should 
have been the case only when the stimulus was the first person of 
another tense. Other observers responded with another verb, and in 
a majority of cases with the same form as the stimulus, but we do not 
know how far this result is due to responses of a phonetic or imitative 
character (as nimmst-gibst, cf. Thumb-Marbe 68, and g.42 below on 
rhyming responses) , which prove nothing for the question of group form- 
ation. Schmidt's results have been tabulated by Thumb (IF 22, 39). 

On the whole, the available evidence is not favourable to Paul's 
assumptions concerning relational (formal) groups. It is true that 
there are some indications that corresponding forms are associated, 
but it is probable that this is due to special factors. Arguments of a general 
character point in the same direction. Basic meanings, being generally 
more palpable and concrete, are much more salient to linguistic feeling 
than the abstract relational meanings. Nevertheless, we have found 
that basic groups are confined to certain definite types, constituted by 
a comparatively limited number of words. It is then improbable that 
identity or similarity of relational meaning should be able to hold 
together groups of the kind assumed by Paul, comprising, for instance,, 
all comparatives, or all nomina actionis, groups that would run into 
hundreds, or even thousands, of words. We have to remember that 
analogy is an all-pervading phenomenon, and if relational groups are 
one of the indispensable foundations of analogy, they must be equally 
all-pervading. The evidence I have been able to bring forward shows, 
at least, that relational groups are not all-pervading. 

9.15. Conclusions. The results of laboratory experiments can 
generally not be applied to speech without some reservations, due to 
the different conditions. Ordinary speech normally stands under the 
control of determining tendencies which guide the trend of thought, 
preserving and strengthening the associations which agree with the 
determining tendencies, but inhibiting irrelevant and disturbing asso- 
ciations (see 5.52). In ordinary speech, therefore, the idea of old will 
normally evoke the idea of new or young only if this association does 
not run counter to the ruling tendencies, or if the control of the latter 
is relaxed through fatigue, deflected attention, or any other reason (cf. 
9.4). The experiments consequently show the conditions in which such 
associations may occur, but not that they actually must occur. This 
argumentation involves the consequence that in ordinary speech analogy 


is rarer than group formation, since the possibilities are made use of 
only in certain circumstances. 

We have found that although basic groups are perhaps sufficiently 
general to play a part in analogical formation, relational groups are not. 
Since Paul's theory, which seems to be the current theory, is based on 
the existence also of relational groups, it is obviously inadequate even 
if no other arguments could be brought against it. 

We have, then, to seek a better foundation for a theory of analogy. 

9.16. The Three Types of Analogy. In proceeding to the explanation 
of semantic analogy, I base myself, as usual, on the linguistic material. 
This results, as shown in the following sections, in the establishing of 
three main types of analogy. 

(i) Combinative analogy, consisting in the isolation and fresh com- 
bination of meanings, basic or relational. 

(2) Correlative analogy, consisting in the naming of a referent with a 
word that is evoked owing to its semantic correlation to another, 
known word, in the same, or in another, language. 

(3) Phonetic interference, consisting in the semantic influence of one 
word on another, owing to phonetic similarity. 

The psychic processes for each of these types will have to be separ- 
ately analysed, since they differ in several respects. 

9.2. Combinative Analogy. 
9.21. Esper's Experiments. The psychic process underlying com- 
binative analogy has been elucidated by some experiments performed 
by Esper, which bring into a convenient form the theories that have 
to be applied (Esper, Exp. Invest.) 

Esper contrived an artificial miniature language. It was based on four figures 
of fantastic shape, designed so as to avoid association with any object of daily 
life. Each figure occurred in four different colours, red, blue, yellow, and green. 
There were thus sixteen figures in all. The observers were told that the experi- 
ment was a test to determine how quickly they were able to learn the names of 
certain sacrificial objects in the Morgavian language, which is spoken on the 
northern slopes of the Himalaya Mountains. The figures were shown one at a 
time, and as a figure was being shown the experimenter spoke its name. After 
the figures had been shown four times (the learning series), they were shown 
again in an entirely different order, and the observers were asked to give the names, 
as far as they could remember them (the recognition series). After that, the 
learning series was again repeated four times, followed by a recognition series, 
and so on, until there had been, on the first day, 32 learning series in all, and 

208 GUSTAF STERN 9. 2 1 

8 recognition series. On the second day there were 16 learning series, on the 
fourth day 8, on the eighth day 4, on the sixteenth day 2, and on the thirty- 
second day I, with recognition series inserted after every four learning series. 
On the sixty-second day there was a final recognition series in order to ascertain 
how much the observers remembered. Esper's main purpose was to study 
contamination between word forms, but analogical formations were also registered. 
In the first experiment, the names of the figures were composed in the following 










nasdeg (65) 








wecdeg (51) 








/owndeg (64) 








rojdeg (66) 



That is to say, the nonsense-words are composed of syllables formed according 
to English speech habits. Each syllable is correlated throughout to one of the 
factors shape and colour. 

In all the learning series, two figures were left out (italicized in the table above). 
They were never shown in the learning series, and their names were never 
pronounced by the experimenter. The observers were thus never taught them. 
In the recognition series, on the other hand, these two figures were included, 
nothing being said to the observers of this circumstance. 

The percentage of correct responses in all the recognition series is given in 
parenthesis in the table. For the figures named it varied between 51 and 75 per 
cent. For the two unlearned figures it was 51 per cent. The unlearned figures 
were thus named as correctly as some of the others. The maximum of correct 
responses in any one recognition series was 86 % for the unlearned figures, and 
88 % for the others. Associative interference (contamination), as for instance 
the response nojling instead of nasling, occurred in a few cases, but no case more 
than twice. 

In Esper's second experiment the figures were the same but the names were 










pelgen (75) 








pegdet (39) 








pezgub (15) 








pembow (36) 



As before, specific sound-sequences are correlated to the two factors of shape 
and colour, but the order of the linguistic elements is shape-colour, thus reversing 
the usual English order of adjective-noun. The linguistic elements are not separ- 
ated by a natural syllable division: nu-lgen, nu-gdet, etc., are contrary to English 
speech habits. 

In this case the rate of learning was much slower than in Exp. I. The 
percentage of correct responses for all the learning series together varied between 
15 and 75 % for the learned figures. The percentage for the two unlearned 


figures was 4 % and 5 %. The maximum of correct responses in any one 
recognition series was 19 % for the unlearned figures, against 61 % for the others. 

The result is thus very different from that of Exp. I. in which the unlearned 
figures were named correctly in 51 % of the responses. There was further a 
strong tendency to contamination, and to a modification of the non-English 
syllable division in accordance with English speech habits. "In fig. i, 2, and 
3, this tendency takes the direction of extending the natural syllables nul-, dojl-, 
and pel-, occurring in nulgen, pelgen, dojlgen}) to the other words. These natural 
syllables thus become semantic units corresponding to shape, while the syllables 
-gen, -det, -gub, -bow, similarly become semantic units corresponding to color" 
(Esper I.e. 38). 

However, Esper concludes that the material in hand is not sufficient to permit 
us to conjecture in what manner the tendencies here apparent with regard to 
contamination would have worked out, or the linguistic system finally have reached 
an equilibrium. I have mentioned this point in passing, and now turn to the 
analogical formations, which are our main concern. 

9.22. Application to Language. We have now to explain how it 
was possible in the first experiment, which is most closely parallel to 
actual linguistic conditions, to supply the names of the unlearned 
figures almost as quickly and correctly as those of the other figures. 

It is clearly a case of formation of categories and general concepts. 
The observers are presented with certain factors running unchanged 
through a series of varying concomitants. In such cases, as is well 
known, the permanent factor may be isolated from its varying surround- 
ings and apprehended as a unit capable of being variously combined 
with other elements. In this case the elements are the syllable nas- 
and the quality of redness, the syllable -kop and a certain shape, and so 
on. This shape, and the quality of redness, are then presented to the 
observer in a new combination, and on condition that the concomitance 
of the two factors with their respective names has been presented a 
sufficient number of times, so that they automatically accompany 
each other in our mind, there will be no difficulty for the observer to 
analyse the new figure into its elements, and to name it according 
to the "language system" of the "Morgavian" dialect, although he has 
never seen the figure before, and has never heard the name he is making 

The process of a combinative analogical formation in real speech is 
exactly similar. A referent is apprehended as being adequately named 
by some known word (stem), and as standing to other referents in a 

^) Note the high percentage of correct responses for these figures. 
Goteb. Hogsk. Arsskr. XXXVIII: i 

210 GUSTAF STERN 9-22 

relation that is suitably denoted by a certain ending (or sometimes left 
undenoted, of. 4.26). Stem and ending are combined in a way that is 
new to the speaker, but which is, nevertheless, felt by him as natural 
and conforming to the rules of his language. The two isolable factors 
are the basic meaning and the relational meaning, as apprehensions of 
the corresponding qualities of the referent, with their respective names. 
Each of these may occur in a variety of combinations, and is thus iso- 
lated and apprehended as a unit, capable of being freely combined with 
other units, within the limits prescribed by the linguistic system. 

In the same way suffixes, like -let, are combined with new items, 
thus extending their range. Esper's words are no doubt to be regarded 
as more closely corresponding to combinations of stem and suffix, or 
of two stems, that is to say, of two basic elements, than to combina- 
tions of basic and relational elements. C. & W. Stern (Kindersprache 
394 sqq.) give interesting lists of childish formations of these types. 

An analogical formation of the present type may be defined as 
being the naming of a previously unknown — at least momentarily 
unknown — combination of a basic and a relational meaning, or of 
two basic meanings, each meaning being expressed by its own 
name, except in cases where the relational meaning is left unexpressed, 
or is expressed by word-order (cf. 4.26). An indispensable condition is 
the existence of one or more previous combinations apprehended as 
such by speakers, and analysed into elements capable of being freely 
combined in various ways, so that the units entering into them are 
available for new formations. Thumb and Marbe have pointed out 
that the number of previous formations serving as patterns for analogy 
or as basis for a scheme, is of subordinate importance. Thus, the Latin 
inchoative element -sc- has derived its meaning from the verb cresco 
(according to Oertel 287. Cf. Bloomfield A J Ph 16 410 sqq., Wundt 

I 445)- 

The basic groups, flexional, derivational, or correlative, are not 
causes of analogy. They furnish the patterns according to which 
analogical formations are fashioned. Series like lead-leader-leading 
occur in great numbers, and it is on the basis of a schematic apprehen- 

1) Cf. Freudenthal 106 sqq. quoting Miiller and Pilzecker § 13. 


sion of such series that a speaker is able to form new nomina agentis 
in -er, or new nomina actionis in -ing from verbal stems. Similarly 
with series like follow- followed, man-man's, swift-swiftly, etc. A new 
formation is not necessarily made on the basis of an individual pat- 
tern. It is more likely to be guided by an abstract scheme, or as Wundt 
(I 461) terms it, paradigmatische Vorstellungsreihen, abstracted from 
a number of groups of one type. The process may, from the point of 
view of its result, be described as a supplementation of basic groups 
or patterns, but that formula gives no indication of the actual psychic 

It is necessary to emphasize the passive, normative part played by 
the groups, the language system, in opposition to the current theory, 
which is represented even by Delacroix: "I'analogie est I'effet de la 
presence dans I'esprit d'un systeme de formes. Elle suppose un mo- 
dele et son imitation reguliere. Elle suppose que les formes se rangent 
dans i'esprit et qu'elles tendent a s'imposer. lya forme analogique 
jaillira sous leur pression." (Delacroix 248). Analogy is not the effect 
of the existence of a system of forms, and the analogical form does 
not issue under the pressure of the form system. On the other hand, 
it is of course quite true that analogy presupposes patterns, and imi- 
tates them. But the effective factor, the causa movens, is the neces- 
sity of filling the functions of speech, and the formal system provides 
only the pattern or norm for the analogical creation. We find again 
that the striving to adapt speech to a better fulfilment of its functions 
is the driving power in the development. 

Another point should be noted. Analogical formations appear in 
the speech of children at a very early period, in normal cases soon after 
the end of the second year of life (W. Stern, Psych. 130, 141, 143). We 
are therefore compelled to explain analogies with the help of psychic 
processes which a child is known to master at that age. According to 
W. Stern's tables (1. c. 143) a normal child begins to use verbs, adjec- 
tives, and particles {i. e., words of relation), and to combine them into 
simple sentences during the latter half of his second year. This implies 
that the child is then able to dissociate activities, qualities, and 
relations from the objects to which they are related. Intentional 
analogical formations do not of course occur at this age. But the 
process assumed by Paul to explain analogies could scarcely be effected 

212 GUSTAF STERN 9.22 

without intentional effort, and is thus altogether out of the question at 
the time when analogies actually begin to appear. (Cf. Stern, Kinder- 
sprache 140 sqq.).^) 

9.23. A Question of Principle. In ME the adjective light acquired the 
meaning 'of small value, cheap' (cf. NED, and Stern, Swift 57). The development 
can be clearly traced. The corresponding meaning of the adverb lightly (Stern, 
Swift 68) is instanced once only, and it does not seem to have arisen through 
independent development of the adverb, but has probably been "borrowed" 
from the adjective. We have to assume that a speaker wishing to express the 
notion of 'cheaply', for some reason selects the stem light-, and, in accordance 
with the usual practice of forming adverbs, employs the word lightly [Me bud 
lihtliche a ping fet me luuep lutel, 'men buy cheaply a thing for which they care 
little'. Ancren Riwle p. 392). 

Theoretically, the word lightly in this phrase may be explained in two ways. 
We can assume that the speaker, forming the meaning 'cheaply', employs the 
well-known word lightly to express it, and feels at liberty to do so, because the 
new meaning of lightly corresponds to a current meaning of the adj. light, and he 
is accustomed to use cognate adjectives and adverbs in corresponding meanings. 
The process is one of group supplementation: the parallelism of meanings between 
light and lightly is completed with a missing member. We have the old word 
lightly used in a new sense. 

Secondly, we can assume that the adverb lightly in this context is a new coining, 

1) We have now found two essential objections to Paul's group theory. First, 
the relational groups do not occur to anything like the extent that would be 
necessary if they really were indispensable for analogical formations; secondly, 
at the time that a child begins to form analogies, he is still unable to perform 
intentionally a complicated operation of the kind involved in Paul's theory. 

A third objection arises from a comparison of the results of Esper's two 
experiments. Why is the process of analogy and learning so much slower in the 
second case? Taken as totals, the words of the second "language" are not more 
difficult than those of the first, and the grouping ought therefore to proceed just 
as easily, with formation of "Proportionengruppen". We should, on Paul's 
theory, expect the results to be about the same in both experiments. On the 
other hand, if we adopt the isolation-and-combination theory, we base it on 
the isolation of elements in the compound words, as names of corresponding ele- 
ments in the referents. Such isolation requires a word as support. The word is 
the centre round which the notion crystallizes, and without a name the formation 
of a general notion is scarcely possible in ordinary circumstances (cf. 3.21, 
5.53, and the writers quoted there). Now, in the second experiment, the 
compounds were formed in such a manner that the second element began with 
a consonant group that cannot be pronounced initially in English: -zg-, -mb-, etc. 
This difficulty retarded the isolation of the phonetic elements, the names, and so 
the formation of the general notion; the analogical process would therefore take 
more time to arrive at a satisfactory result. 


which the speaker makes on the basis of the adjective light, but without associating 
it with the previously — in other meanings — current adverb lightly. Genetically, 
we have then not the old word in a new meaning, but really a new word, although 
identical in form with the old word; a new word created through combinative analogy. 

A similar problem arises with regard to shortenings. Wellander (Studien II 
3 sqq.) contends that the German word Schirm 'umbrella', is a shortening of the 
full term Regenschirm and not a specialized meaning of the general term Schirm 
'shelter'. From a genetic point of view, Schirm 'umbrella' is not the old word 
with a changed (restricted, specialized) meaning, but a new word. I have 
objected (Stern, Litteris III 54) that it is not always so. If a type-writer is called 
the machine, we can explain it only as a specialized use of the generic term. Why 
may we not explain Schirm 'umbrella' in the same way? It is true that in the 
latter case the generic name enters into the full name of the species as its second 
element, but that is a mere coincidence (See below 10.412). 

Returning to the analogical formations, there are cases in which the second 
explanation is alone possible, viz., those in which the word coined by analogy 
did not previously exist. But with regard to lightly and similar words, both 
explanations are possible, just as Schirm 'umbrella' might be an instance of 
specialization or of shortening. Which is the correct explanation? 

I do not think it is unfair to counter this question with another: does it matter? 
For if we regard lightly as a new coining, we must assume (i) that the speaker 
who makes it and repeats it cannot in the long run avoid associating his lightly 
with the current word lightly, and thus apprehending them as the same word; 
and (2) that the hearer must at once apprehend lightly as being the current word, 
which, with the help of context, he is able to interpret in a new but perfectly 
natural manner, the new meaning corresponding to a familiar meaning of the 
adjective light (cf. Hatzfeld, Bedeutungsversch. 53). 

For the speaker it is, then, for all practical purposes indifferent if lightly 
'cheaply' is a new coining which is, possibly at once, associated to the current 
word lightly, or rather identified with it; or if lightly is the current word itself with 
a new meaning. To separate the two would be to make a distinction without 
a difference. For the hearer, the process is in both cases theoretically and actually 
the same. 

We may adduce in this connection the fact, pointed out by Paul and others 
with regard to morphological analogies, that "es ist fiir die Natur dieses Vorganges 
ganz gleichgiiltig, ob dabei etwas herauskommt, was schon friiher in der Sprache 
iiblich gewesen ist, oder etwas vorher nicht Dagewesenes" (Paul, Prinz. no). 
A speaker does not learn all speech forms directly, but coins many of them by 
analogy. Children often do this in such a way as to be corrected, but when they 
make correct forms, these are accepted and added to their vocabulary. In a 
similar manner we must assume that even if a certain meaning of, say, an adverb, 
is current in the language, one or more speakers may not have heard of it, but 
nevertheless make use of it analogically, because they know that the cognate 
adjective has a corresponding meaning (cf. Paul, 1. c, Jespersen, Language 128, 
163, Oertel 163 sqq.). 

214 GUSTAF STERN 9.23 

Paul remarks (Prmz.195): "Eine Bedeutungserweiterung des Grundwortes 
Oder des dem Sprachgefiilil als solches erscheinenden Wortes teilt sich leichter der 
Ableitung mit, als umgekehrt eine Bedeutungserweiterung der Ableitung dem 
Grundwort." This is easily explicable on the combination theory: when we form 
a derivation we use the stem itself, and any meaning attaching to it can easily 
arise in mind along with it. A new meaning acquired by a derivation, on the 
other hand, is felt as belonging primarily to the derivation only, and is not so 
easily called up in mind in connection with the stem, if not specially induced by 
the context. 

With regard to the psychic process in the speaker's mind when using lightly 
'cheaply', I analyse it in accordance with the theory illustrated by Esper. The 
referent is analysed by the speaker as having the quality of cheap-ness; this is 
the referent of the basic meaning, which is expressed by the stem light. The 
quality is attributed to a verb, and is consequently adverbial in character; this 
furnishes the relational referent, which is denoted by the ending -ly. 

The driving power for the innovation is, as usual, the functions of speech and 
the immediate purpose of the speaker. The derivational groups of the relevant 
kind (adjective-adverb) furnish a scheme which acts as a norm, determining the 
form of the analogy. For the purposes of the present study I take up a purely 
empirical position. We find that lightly occurs in the sense of 'cheaply', which 
does not seem to be explicable by an independent development of the adverb 
itself, but which corresponds to the sense 'cheap' in the adjective light. Whether 
lightly in its new meaning is also morphologically a new coining, or if it is the 
old word with a new meaning given to it — that is a question which I leave aside 
as irrelevant. The meaning is, in whatever way we explain its connection with 
the word, due to an analogical process, a supplementation of the derivational 
scheme, and I class it as such. The result of this process is a "sense-loan" frotn 
a cognate word, or another flexional form of the same word. With regard to 
the psychic mechanism the type is altogether different from what is usually 
termed sense-loans, from foreign languages (see 9.33). 

I shall now proceed to discuss the different types of combinative 
analogy, and shall distinguish analogy in flexional, in derivational, and 
in correlative groups. 

9.24. Flexional Groups. An analogical sense-change takes place 
in flexional groups as a supplementation of the flexional pattern when 
a member of the group acquires, otherwise than through independent 
development, a meaning that did formerly not belong to it, but which 
corresponds to a meaning belonging to some other member or mem- 
bers of the group. 

The OE adverbial comparative ra^er signified 'sooner, more quickly'; 
but that which happens 'sooner' than another event or action, also 
happens 'earlier': we find equivocal instances in OE: Hwi comon ge 
raj)ur ponne eower gewuna wcss 'why did you come more quickly, or 


sooner, or earlier, than usual?' (^Ifric, Exodus 2.18), which shade 
into the fully developed meaning 'earlier': sume lator felad pare IcBce- 
doma, sume ra^or, 'some feel the medicine later, some earlier' (lysece- 
boc 26.36); the sense is well instanced in ME: A hwilke time se eure 
Mon of pinchp his mis-dede, O^er ra^er oder later, mike he seal imeten 
(Poema Morale lyamb. 131). In the positive, we do not find the sense 
'early' until the ME period (earliest instance: Al pat pu singst rape 
oper late, Owl and Nightingale 1147), and there are no intermediate 
instances between 'soon' and 'early'. It seems probable that the 
change took place in the comparative only, and that the new 
meaning was afterwards extended to the positive (Stern, Swift 34 sqq.). 

It is of course usual for the positive and the comparative forms to 
have perfectly corresponding meanings. By the development of a 
new meaning in the comparative, a gap is caused in the group, and it 
is filled by supplementing the scheme with a corresponding meaning 
for the positive. As soon as the need, or the desire, to do so arose, 
this supplementation could take place. 

Judging by the instances given in the NED — I have no other 
evidence for this word — a similar process has taken place with regard 
to soon 'readily, willingly'. For the comparative, the sense 'more 
readily, easily' is quoted from the Ancren Riwle (a. 1225), and the 
sense 'more readily as a matter of choice', which apparently arose out 
of the former, from Hardyng's Chronicle (1458). The positive as soon 
as 'as readily as, as willingly as', is not quoted till Shakespeare, and 
it appears that the sense was introduced from the comparative into 
the positive. 

Some Latin instances are given by Hey. The participle suspectus 
has acquired the meaning 'suspected', probably owing to the influence 
of the noun suspicio. The verb suspicere, on the other hand, retained 
only the meaning 'to look up to, revere'. But in Sallustius we find: 
suspectus regi et ipse ettm suspiciens novas res cupere. In this connection 
suspicere has been influenced by the preceding suspectus and has re- 
ceived the meaning 'to suspect'. A levelling out in the opposite direction 
is instanced by Juvenalis 9.57: te Trifolinus ager fecundis vitibus implet, 
suspectumque iugum Cumis et Gaurus inanis 'the mountain ridge looked 
up to by Cumae' (Hey Ahh 13, 214). 

Sallustius once wrote aquis hiemantibus 'the stormy waters'. But 
in the sense 'to storm' hiemare is regularly only impersonal, while in 

2l6 GUSTAF STERN 9.24 

the sense 'to hibernate' it has also personal forms. Sallustius has 
coined a new form of the impersonal verb, perhaps influenced by the 
personal forms of the other meaning (Hey 1. c). — I give these in- 
stances on Hey's authority. 

In most cases, sense-changes take place in such circumstances that 
any flexional form of the word may be equally affected. It is only 
exceptionally that a particular expression, requiring a particular 
flexional form, is involved, as in the instances quoted. Moreover, 
it requires a detailed knowledge of the history of the word to enable 
us to ascertain the precise nature of the process, and consequently 
good instances of this type are very rare. 

9.25. Derivational Groups. The analogical process in supplement- 
ing derivational schemes has attracted more attention and is better 
known than that treated in the previous paragraph (cf. Hatzfeld, 
Bedeutungsversch. 34 sqq., 46 sqq., Delacroix 207). The following 
instance is typical (Stern, Swift 108 sqq.). 

The ME. adverb jaste originally signified 'firmly, immovably'. 
By a development that can be traced in detail the adverb acquired 
the meaning 'vigorously, violently, eagerly', and further that of 
'swiftly', a stage which was reached about 1300. The cognate adjective 
fast originally' signified 'firm', immovable' (passive strength). It 
took no part in the development of the active sense in the adverb, and 
there are no traces of a meaning 'vigorous, violent, eager'. Towards 
the end of the 14th century, we find a few instances of the meaning 
'swift', which is now a common one in the adjective: a vast vleynge, 
a faste trott, no faster course, 'a rapid flight, a rapid trot, no swifter 
course'. As stated in the NED, this is clearly a case of borrowing from 
the adverb (cf. Stern, 1. c). 

The psychic process is identical in principle with that of flexional 
supplementation. I refer to the analysis in 9.23. Note that the 
earliest instances of the adj. fast 'rapid' are all of them qualifiers of 
nomina actionis, of which two, vleynge and trott, correspond to common 
verbs of motion. Since the adv. faste 'rapidly' was common, the 
phrases flen faste, trotten faste (to flee, to trot rapidly) must have been 
well-known to speakers, and perhaps served to facilitate the change. 
In other cases, no such intermediary phrases are found, and it is 
uncertain if they are necessary. 

Note further that it is only the meaning 'rapidly' that is borrowed 


by the adjective. The adverb faste was, at one period of ME, very 
common in the senses 'vigorously, violently, eagerly', with an intensi- 
fying force, but this sense, although much more common than that of 
'rapidly', is not borrowed by the adjective. One would think that 
just as it is easy to go from faste flen to a fast fleing, it would be easy 
to go from faste fighten 'to fight vigorously', to a fast fight 'a vigorous 
or energetic fight', but there is no trace of such a meaning of the 
adjective. The reasons for this are entirely obscure. 

Like fast, the adjective swi^ 'strong, vigorous, forcible', has 
acquired the sense of 'swift, rapid' by borrowing from the correspond- 
ing adverb, swipe, in which that sense originally developed (Stern, 
Swift 127). 

The adj. hard acquired the sense 'difficult to do or accomplish, not 
easy' in early ME (Ormulum, 1200, NED). In the adverb hardly, the 
corresponding sense is not found (according to the NED) till Cover- 
dale 1535. Even if somewhat earlier instances should be discovered, 
it seems reasonable to assume that the meaning in question was trans- 
ferred from the adjective to the adverb. In the adverb there then 
arose, through independent development from the sense mentioned, a 
new meaning, 'not quite, scarcely'. This is instanced from 1553 and 
there is no corresponding sense in the adjective. 

Reversion belongs to revert, and the proper noun to reverse is reversal. 
But the greater frequency of the ending -ion has led to reversion being 
used as noun to reverse. (Cf. Fowler, MEU: "the reversion of our Free 
Trade policy"). 

The members of a derivational group may be differentiated by 
their prefixes. ly. ascendere and descendere must be assumed to form 
a group by reason of their common stem and correlated meanings. 
In this parallelism there was one exception: descendere signified also 
'to be a descendant of, while ascendere could not mean 'to be an ances- 
tor of. But we find ascendentes used as meaning 'ancestors in a direct 
line', just as descendentes means 'descendants in a direct line'. The 
two words must have been analysed by speakers as compounds, in 
which the prefix indicates the direction of the relationship denoted 
by the stem. When a speaker sought for a word denoting ancestors, 
he would be brought by his acquaintanc-e with descendentes to analyse 
the notion as consisting of an idea of direction (motion) and an idea 
of relationship, and consequently employed ascendentes. — Since the 

2l8 GUSTAF STERN 9.25 

meanings of the two words stand to each other in a relation of oppo- 
sition, it might be possible to explain the analogy as correlative. 

Other lyatin instances are adduced by Hey (ALL 13, 211 sqq.). 
Or are acquired the meaning 'to pray', and or alio borrowed from it 
the corresponding meaning 'prayer'. Truncus primarily denoted the 
trunk of a tree. The sense 'fragment, cut off piece' is from the verb 
truncare 'to make into a trunk, to cut off, to mutilate'. Praedo 'rob- 
ber' is once found in the sense of hunter, owing to the influence of 
praeda 'quarry, prey'. (Cf, Paul, Prinz. 192 sqq., 244 sqq.). — Fr. 
bouchon signified (i) stopper, and (2) handful of straw. The verb 
bouchonner had the two corresponding senses. The second of these, 
'to rub with straw', developed the metaphorical meaning 'to caress, 
to stroke', and then the noun analogically acquired the sense 'darling', 
of little children. (Hatzfeld, Bed.-versch. 47, quoting Diet. Gen.). 

If the supplementation leads to the creation of a form not previously 
existing in language, the process is the same but we have no sense- 
change. Thus when the adjective moony was coined, probably on the 
analogy of starry, sunny, etc. Words of this type fall outside the scope 
of my investigation; they are of course very numerous. 

As new words I regard also the use of, for instance, iron as an 
adjective or as a verb. The change is mainly one of relational meaning, 
and as such it is excluded here. The process is extremely common in 

We might call this type "sense-loans from cognate words in the 
same language", but, as will be shown in the next section, the psychic 
process in sense-loans from other languages is quite different. 

9.3. Correlative Analogy. 
9.31. The Psychic Process. The basis of this type of analogy is 
the correlative group formed by two or more words. I may take as a 
typical instance the group formed by two corresponding words in two 
different languages, as French arriver and English to arrive. Their 
meanings correspond on most points, and to persons knowing both 
languages the two words with their meanings (the two flexional groups) 
form together a correlative group, the relation holding between them 
being one of similarity. There is, however, one gap in the correlation 
of meanings: arriver means also 'to attain success', and this meaning 
was originally strange to the English verb; it has now made its appear- 


ance. It is not given by the N^D, but is found in the COD, and in 
Fowler, MEU, as a Gallicism. 

I explain the process in the following manner. An English speaker 
with a knowledge of French has occasion to denote an action (the 
referent) which he apprehends as 'attain success'. For some reason 
— perhaps recent use — the French word arises in his mind. In 
seeking an English equivalent, he is led by the phonetic similarity 
and the accustomed correlation between the meanings of arriver and 
arrive to think of the latter, and finding no obstacle to its use, he em- 
ploys it. I think that the new meaning should not be too far away 
from the traditional range of the word in question. In the instance 
given, arrive can easily be understood as metaphorical. 

The starting-point for the process of analogy is the French word 
with its meaning; the relation of similarity presents itself as a means 
of solution, since what is sought is an English word to correspond 
with arriver. We have thus one fundament and a relation, and the 
second fundament automatically presents itself in the shape of to 
arrive (see on the eduction of correlates 4.265). 

9.32. Within one Language. Words forming correlative groups 
often influence each other's form (cf. 9.13). It is therefore natural 
to assume that they also influence each other's meaning, but reliable 
instances are rare. A very detailed investigation of the history of a 
word is required to show up such influences. 

Black letter day is 'an inauspicious day; as distinguished from a red 
letter (or auspicious) day; the reference being to the old custom of 
marking the saints'-days in the calendar with red letters'. Black letter 
has here, through analogy, received a meaning which is otherwise 
unknown in the combination, the ordinary meaning being that of 
'Gothic' or 'Old English' type. High signifies 'intense, extreme' and 
was used in that sense of opinions, doctrines, and so forth. We have 
it in the phrase High Church. It seems not improbable that the 
corresponding phrase Low Church is due to analogy, although 
the chronological difference between the two earliest quotations in 
the NED is insignificant (cf. NED s. v. high adj. 15 b). 

Hey states that in lyatin, where equites peditesque was a current 
phrase, equites became the designation of a social class, the knights. 
This led to pedites receiving the meaning 'common man, common 
citizen' (Hey, ALIy 13, 220). It seems to me that other explanations 

220 GUSTAF STERN 9.32 

might be imagined. Oertel (166) states that "when constantia was 
used metaphorically to denote a certain mental quality, it "induced" 
a similar change in its opposite mohilitas" . 

9.33. Two Languages Involved {Sense-loans). 

9.331. Theoretical Discussion. In the instance given above 
(arrive) there was not only a similarity of meaning but also a similarity 
of form between the two words involved, and it would seem that in 
the case of unintentional sense-loans formal similarity is of great im- 

Sense-loans on a large scale occur among bilingual speakers. If we 
were able to trace in detail the semantic history of English from 800 
to 1 100 we should probably find that the meaning of many English, 
words had been influenced by their Scandinavian correspondences. 
When eorl, 'earl', originally 'a man of noble rank', as distinguished 
from a ceorl, 'churl, or ordinary freeman', assumed the sense of 'the 
vice-roy or governor of one of the great divisions of England', this is 
clearly due to the fact that earl was used to render Scand. jarl, in the 
place of native English alderman. I^ater on, earl was applied t& 
'all feudal nobles and princes bearing the Romanic title of Count ' 
(NED), which we may regard as a second sense-loan. In the first 
instance, earl renders a cognate word of great phonetic similarity, in 
the second instance it renders a totally different word. lyacking the 
necessary material it is impossible to say at present whether these 
two sense-loans are both unintentional, or whether the second is inten- 
tional; i. e., if the phonetic similarity is necessary for unintentional 
sense-loans. The instances from OE and ME adduced in Stern, Swift, 
are all of them found in translations or else in texts obviously influ- 
enced by French or L,atin models, and they are therefore to be regarded 
as more or less intentional. 

In accordance with my general principles I leave aside cases in 
which a foreign word is adopted,^) and discuss only instances of the 

^) I exclude also the coining of a new word of native material in order to render 
a foreign word: as bodeful for ominous, folklore for tradition, birdlore for ornithology, 
foreword for preface, betterment for improvement, and the innumerable German 
coinings of the same kind: Jungfernrede for maiden speech, Halbwelt for demi- 
monde, etc. See Sailer ZsfddU 31, and Fowler, MEU s. v. Saxonisms; see also 
1 1.2 below. 


type already given: a foreign word is rendered by a native word which 
in the main corresponds to it, but has not previously been used in the 
sense now occurring. Several authors (see for instance Wellander, 
Studien I 103) speak of foreign influence on the "inner speech-form" 
of a language, but I see no necessity for introducing the term here. 

With regard to points of view for the further classification of sense- 
loans, I have already mentioned the distinction between intentional 
and unintentional, which, however, is inapplicable owing to insufficient 
material. Another distinction would be that between cases in which 
the foreign and the native word are similar (as earl and jarl), and those 
in which they are not (as earl and count). But we cannot say if this 
distinction is essential or not. Seiler's distinction between words in 
which the earlier native meaning has died out, and only the new bor- 
rowed meaning remains, and words in which both remain, seems unessen- 
tial with regard to the loan itself (Seller, ZsfddU 31, 241 sqq. See 
Wellander, Studien I 103 for earlier classifications). 

Sandf eld- Jensen (Caiques 166 sqq.; see also the literature quoted 
there) makes a threefold distinction, (i) lyC sens d'un mot s'elargit 
d'apres les significations du mot correspondant d'une autre langue 
(emprunt semantique). Earl and arrive are instances of this type. 
(2) La traduction sert a la formation de mots nouveaux: G. entdecken 
is coined on the pattern of F. decouvrir, F. surhomme and E. superman 
on the pattern of G. Uebermensch. This type, as involving the coining 
of new words, may be left aside. (3) On traduit des tournures de 
phrase et des locutions. G. den Hof machen is a copy of F. faire la cour, 
'B. to leap to the eyes, of F. sauter aux yeux. Seller (1. c.) makes a 
similar distinction between Bedeutungsentlehnung and Lehnredens- 
arten. As a more formal criterion, it may be useful in dealing 
with the material. In actual speech all words occur in contexts, and 
the difference between the first and third of Sandfeld-Jensen's types 
is the fact that in the former the change affects one word in a momen- 
tary combination, in the latter type it affects one or more words in a 
permanent combination, a standing phrase. Since no other classi- 
fication is available, I shall make use of it, and also distinguish as far 
as possible between loans that are influenced by phonetic similarity, 
and loans that are not so influenced. 

9.332. Single Words, Semantic Similarity only. In OE and ME 
the influence of Latin and French was often the cause of sense- 

222 GUSTAF STERN 9-332 

changes in the English words. OE cwic 'living' had many senses in 
common with Latin vivus, and was therefore used to render it in panis 
vivus, aqua viva, and other expressions introduced by Christianity 
(Stern, Swift 151, 161). Similar changes on a large scale often take 
place when a cultural movement is introduced into one country from 
another in which it has already developed a technical vocabulary. 
(See Carnoy 135, 229, Sandfeld-Jensen, Caiques 167, Bally I 48, 
lychmann, Bedeutungswandel, Seiler 1. c, Wellander, Studien I 103 
sqq., Paul, Prinz. 401 sqq.) 

OE Me leoht sleep oferarn 'a light sleep fell over me', is a literal 
translation of levis mihi somnus obrepsisset (Bede, Hist. Eccl. 410. 12). 
With the exception of one instance in the Prose Guthlac, which is 
probably also due to Latin influence, I have found no instance of the 
phrase light sleep in OE or ME, and the earliest quotation in the NED 
is from 1827 (Stern, Swift 52). Chaucer writes. Never ne shal his hy~ 
tinge bisinesse forleten him whyl he liveth, ne- the lighte richesses ne 
sholle nat heren him companye whan he is ded, where hytinge translates 
mordax, and the last phrase runs in the original, defunctumque leves 
non comitantur opes (Chaucer's Boethius B. Ill M.3:6. Stern, Swift 
53. See also ib. 276). 

According to Oertel (167) Mrs. Humphry Ward uses overdrive in 
the sense of German iibertreiben 'exaggerate', when she says overdriven 
realism, you overdrive your duties. 

Exposition is used in the sense of 'exhibition' under influence of F. 
exposition (Fowler, MEU s. v. Gallicisms).^) 

German handbooks give many instances of this type. G. Geschmack 
follows the meanings of F. gout, and Kunst those of L. ars and F. art, 
Schonheit those of F. beaute. (Seiler 244). G. Ueberzeugung originally 

1) As exemplifying the difficulties which attend the study of this type I shall 
quote the remarks of the NED on the sense-development of the prep, of, which 
was influenced by Latin and French: "Even in OE., this native development was 
affected by the translational character of the literature, and the employment of 
of to render L. ab, de, or ex, in constructions where the native idiom would not 
have used it. Of far greater moment was its employment from the nth c. as the 
equivalent of P. de, itself of composite origin, since it not merely represented L. 
de in its various prepositional uses, but had come to be the Common Romanic,, 
and so the French, substitute for the Latin genitive case. Whether of might 
have come independently in Eng. to be a substitute for the genitive is doubtful . . . 
but the great intrusion of of upon the old domain of the genitive, which speedily 


'act of proving some one guilty by means of witnesses' corresponded 
to F. conviction. The latter word acquired the sense 'persuasion', 
and this was borrowed by the German word (Hey, ALL 9). 

9.333. Single Words, Semantic and Phonetic Similarity. In addi- 
tion to arrive (9.331), Fowler (MEU, s. v. Gallicisms) quotes as instan- 
ces of "giving to an existent English word a sense that belongs to it 
only in French, or to its French form only", intrigue, verb, 'interest, 
perplex', impayable 'priceless for absurdity, impudence, etc.', actual 
'concerned with the present', as in: the most actual and instructive 
article is on broadcasting". It seems difficult to distinguish these in- 
stances from Fowler's fourth type: "substituting a French form or 
word that happens to be English also, but in another sense, for the 
really corresponding English", as brave for honest or worthy, or ascen- 
sion for ascent. 

Novel, subst., shows repeated sense-loans. It was first brought into 
English from the French, in the two senses 'something new, a novelty', 
and 'news, tidings' (1460 and 1475, respectively, in the NED). In 
1566 it is found in the sense of 'tale or short story', due to It. novella, 
Sp. novela, used of the stories in Boccaccio's Decamerone, and others 
of that kind. Finally, it was employed to designate 'a new decree or 
constitution, supplementary to the Codex', which was an adaption of 
Lat. novella (sc. constitutio) . The sense of 'judicial or legislative assem- 
bly' in E. chamber, is borrowed from L,. camera and F. chambre; simil- 
arly G. Kammer and Swed. kammare. 

E. slim 'sty, cunning' is said to be an adoption from S. African Dutch, 
at the time of the Boer war. F. selection has taken on technical senses 
after E. selection (Darwin); similarly Swed. urval (Darmesteter, Mots 
nouveaux 61; further Hatzfeld Bed.-versch. 5; Carnoy 230; Wellander, 
Studien I 128). 

extended to the supersession of the OE. genitive after adjectives, verbs, and 
even substantives, was mainly due to the influence of F. de. Beside this — the most 
far-reaching fact in the sense-history of of ■ — the same influence is also manifest 
in numerous phraseological uses, and esp. in the use of of = F. de in the construction 
of many verbs and adjs. Many of these can be clearly distinguished; but, in other 
cases, the uses derived from F. de have so blended with those derived from OE. 
of, giving rise again to later uses related to both, that it would be difficult, if not 
impossible, to separate the two streams, with their many ramifications". This 
quotation clearly shows what careful research would be necessary to disentangle 
all the threads in the sense-history of the "big" words. 

224. GUSTAF STERN 9-334 

9.334. Phrases. It is very common for whole phrases of two or 
more members to be taken over, and this may easily lead to one or 
more words receiving a new meaning. Wellander points out (Studien 
I 123) that when G. Geist adapts itself to the various meanings of F. 
esprit, this is due to a series of literal renderings of phrases like Geist 
des Jahrhunderts from esprit du siecle, Unternehmungsgeist from 
esprit d'entreprise, schoner Geist or Schongeist from bel esprit, and so 
on. The second element is isolated from these combinations in its 
new shade of meaning, and is then capable of being used alone in the 
new sense. Similarly German Vorlesungen iiher Geographie was made 
into Swedish jorelasningar over geografi, and the particle over thus 
received a previously unrecorded meaning of 'on, concerning', in which 
sense it was isolated and entered into new combinations, without 
German prototypes. 

Among such caiques linguistiqiies in English, Fowler quotes, from 
French, to jump or leap to the eyes, to the foot of the letter, give furiously 
to think, knight of indtistry, daughter of joy, gilded youth, living pictures 
[tableaux vivants), the half-world, rose-colour, curtain-raiser, do one's 
possible, castle in Spain, goes without saying, suspicion (soupgon), dean 
{doyen), marriage of convenience, on the carpet, success of esteem. The 
phrases are generally intentional, and the ensuing sense-change prob- 
ably largely contextual. 

To give a person the sack is a copy of a French phrase, current in Fr. 
from the 17th century: On luy a donne son sac, 'hee hath his pasport 
giuen him' (said of a seruant whom his master hath put away; Cot- 
grave). Cf. Dutch iemand den zak geven, to give one the sack (already in 
MDu.), den sak krijgen, to get the sack (NED). I do not know if Eng- 
lish linguistic feeling really associates this locution with the old noun 
sack 'bag'. 

9.34. The Pfoblem of Synonyms. Many writers have assumed that synonyms 
influence each other in meaning, and also, to explain the influence, that synonyms 
form associative groups. These would then be a kind of correlative group. Hey 
(ALL 13, 218) speaks of "jene gewohnliche und allgemeine Beeinflussung der 
Synonyma untereinander, vermoge deren die Grenzen im Gebrauche sich allmah- 
lich verwischen". Falk (Betyd. 99) states that similarity of meaning may be the 
sole effective factor in abolishing the difference of meaning between words of 
similar import: partial synonyms become total synonyms. i) 

^) Falk gives the two following instances: Norw. fiff (Germ. Pfiff 'trick') has 
acquired the secondary meaning of 'finery' through influence from Norw. puss 


But it is well-known that total synonyms are rare. If it was true that synonyms 
influence each other in the direction of a levelling out of differences, we should, 
in view of the very long time during which this tendency must be assumed to 
have been active, expect to find a large number of total synonyms in every 
language. As we do not, the levelling-out tendency must be weak or non-existent, 
or it must meet an opposition that is even stronger (cf. Beckman 123 sqq.). 

Falk further says that many notions are in ON represented by a number of 
words which no doubt primarily possessed differing shades of meaning, but finally 
fell together, whereafter generally one word supplanted all the others. Some of 
Falk's instances (1. c. 48 sqq.) are of the following type: "By the side of the prosaic 
words for 'fight', orrosta and vig, there were the poetic bod, gunnr, hildr, hjaldr, 
imun, rimma, rosta, snerra, Prima, orlygi, etc., and we do not know if there 
was any difference of meaning between them. During the course of linguistic 
development the words that were restricted to the poetic language have generally 
been ousted in the competition, wherethrough our memory has been delivered 
from many burdens" (I have translated Falk's text literally). 

It seems to me quite natural that words belonging to an artificial dialect like 
the ON. poetic language, and not to popular spoken language, should disappear 
with the disappearance of the old poetic tradition. Where is there any trace of 
analogical influence in this process? And if all the words quoted above were 
at last really used promiscuously, it seems highly improbable that this was the 
result of mutual analogical attraction. It is much more likely to have been 
an instance of linguistic and objective uncertainty (cf. 4.17), the result of the 
disappearance of the old tradition concerning various conceptions connected with 
fighting, of a profound change in the technique of fighting, and in the outlook 
on life, during which the linguistic tradition of poetry also disappeared. 

Kroesch (Analogy 39) formulates the current opinion in general terms. "A 
word X with a meaning A, develops from this a meaning B. Thereupon a word 
Y, also with the meaning A, a synonym, being associated with X, likewise deve- 
lops the meaning B", the reason being the influence of X on Y, based on the 
association of the two words. Among the instances adduced by Kroecch, I may 
quote French chiquer 'beat', which develops the meaning 'deceive' and so do like- 
wise torcher, taper, estamper, toquer, craquer, aquiger (Fr. attiger), which also meant 
'beat' (quoted from Schwob & Guieysse 49 sqq.). 

There are several points to be noticed in this connection. First, as I have 
previously shown, the existence of a group is not the cause of an analogy, but 
merely provides a pattern. The function of speech is the cause. If we assume 
that a speaker wants to denote the referent P, which is often denoted by the 

(Germ. Putz and Posse) which has both meanings. But according to Hellquist, 
Ft. Ordbok, it is not certain that fiff in the latter sense is the same word. Could 
it be echoic? Falk further states that Norw. fundere 'to found' (L. fundare) has 
developed the meaning 'to ponder' through the influence of Norw. grunne, which 
has both meanings. But according to Hellquist, quoting O. v. Friesen, fundere 
'to ponder' is derived from a Germanic stem, and has nothing to do with fundare. 
— The instances are therefore not convincing. 

Goteb. Hogsk. Arsskr. XXXVIII: i. 

226 GUSTAF STERN 9.34 

word X, meaning B, and also having the meaning A, at the moment irrelevant, 
his thoughts, in order that the analogy may occur, shotdd in some way be led 
to the word Y, which means A, but not B — I have retained the symbols em- 
ployed by Kroesch and assumed exactly the same situation as he does. But 
when a speaker wants to denote the referent P, and apprehends it as B, and 
denotes it by X, there is no reason why the thought of the irrelevant meaning 
A, and its referent Q, and its potential expression Y, should enter his mind at 
all. And I do not see that this fact can be altered by the assumption of any 
other plausible situation. 

I see only one possible exception to this rule: if the two words in question 
are in any way specially connected, as for instance if they ate members of the 
same derivational group, but then the case falls under the heading of combinative 
analogy, and the psychic process is quite different. 

With regard to the French argot-words adduced by Kroesch, it is expressly 
stated by Schwob and Guieysse that the intentional use of synonyms for the pur- 
pose of intensification and for making speech less easily intelligible is peculiar 
to argot. Such intentional, and therefore metaphorical or hyperbolical use of 
synonyms is something very different from analogy, and falls under the heading 
of Nomination (ch. 11). 

However, if we turn to the core of the matter, there is no doubt about the 
existence of sets of parallel sense-developments, groups of synonyms which follow 
the same, or approximately the same lines. Thus, in 7.45 above, I have given a 
list of 23 English adverbs signifying 'rapidly', all of which also develop the sense 
'immediately'. Should we explain this parallelism by analogy? 

Analogy presupposes group formation. The first question is therefore: do 
such synonyms form a group? 

The psychological evidence is, as far as I have been able to discover, very 
meagre. I have only found one statement on the problem, by Jung-Riklin (III 
72), and to the effect that synonym responses were rare in their experiments. The 
tables given by the writers quoted in the present chapter have little to show 
in the way of synonym reactions. 

The contamination between synonyms, occurring in its most extreme and 
intentional form in the so-caUed portmanteau-words, has been adduced as a proof 
of association between synonyms; but this is, I think, a mistake. If a pair of 
synonyms are confused by a speaker, this does not show that the two words were 
previously associated with each other, but only that both are associated with the 
referent that the speaker wants to denote, so that the thought of the referent 
may call up either word. That is nothing new, and proves nothing. 

It has already been pointed out that total synonyms are rare. Synonyms may 
be defined as words with identical or partly identical referential range, but 
different semantic ranges. That is to say, they denote the same referents, but 
each word denotes it in an aspect that somehow differs from the others. When 
a speaker wants to denote a referent, he is practically always seeing it in a 


peculiar context, into which one of the synonyms may fit, but not necessarily 
the others. There is then no reason for these to arise in his mind.^) 

The reason for the parallelism of sense-changes has to be sought in another 
quarter. If we knew nothing about the actual history of the adverbs just 
mentioned (7.45) the series might be regarded as an excellent instance of the 
analogy hypothesis. But on a closer study of the words we find that an indispens- 
able condition for the change from 'rapidly' to 'immediately' is the use of the 
adverbs in a definite context, and we find that this context recurs in every 
single case. If the change was due to analogy, that would not be necessary: the 
association with the synonyms would of course exert its influence equally in any 
context. We conclude, then, that parallel developments are due to similar 
circumstances, just as parallel phonetic changes affect all sounds of one kind in 
a specific position. It is natural that synonyms should be used in similar contexts, 
which are the expressions of similar mental content, and denote similar referents. 
It is then to be expected that the semantic consequences should be similar. 

Three more observations may be made. If the cause of a change like that of 
'rapidly' to 'immediately' could be analogy, it is reasonable to assume that the 
tendency would at least not grow weaker as more and more words acquire both 
meanings. But we find that the tendency to the change ceases altogether at 
some time during the late Middle Ages. This is inexplicable if analogy is a 
strong factor in semantic change. 

Further, if the fact that the word X has the meanings A and B, makes the 
word Y, with the meaning A, assume also the meaning B, would it not also 
make words with the meaning B assume the meaning A? In other words, is 
there any reason why the analogy should not work both ways? Returning, again, 
to our English adverbs, we find the remarkable fact that there is not a single 
instance of a change from 'immediately' to 'rapidly'; and the same state of things 
has been found to exist in German (Stem, Swift 218 sqq., and Sandegren 106). 

Finally, if synonyms form associative groups, and therefore influence each 
others' meanings, it is reasonable to assume that they should also influence each 
others' form, gender, etc. as do the members of other correlative groups. I have 
quoted in 9.13 a number of such groups, and the additional material available 
in the various handbooks is considerable in quantity; in it synonyms are 
conspicuously rare. This is inexplicable if they form groups. But if they do 
not, i. e., if they are not especially associated with each other, how could they 
influence each others' meanings? 

I think I am on firm ground in questioning the assertions of Kroesch and other 
writers. A bare list of parallel semantic developments, plus an assertion that 
the parallelism is due to analogy, is not evidence. There is no trace of analogical 
influence in the series of adverbs which I have quoted, in spite of the fact that 
they seem to offer an ideal ground for such activities — if the current assumptions 
were correct. Until more evidence is brought forward I think I may leave the 
question of synonym groups aside, as not proven. 

*) Anyone who has worked at cross-word puzzles can testify that synonyms 
are often very difficult to recall. 

228 GUSTAF STERN 9.35 

d.35. Wundt's Korrelative Laut- und Bedeutungsanderungen. Wundt speaks 
of correlative phonetic and semantic changes, by which he means cases in which 
a word is differentiated phonetically into two forms, to which different meanings 
attach themselves — so-called doublets. The phonetic differentiation is, according 
to Wundt, always the prior process. If a word A splits into two, B and C, and 
we denote the meanings of these by a, /?, y, then the semantic differentiation 
into a, /?, 7", is always posterior to the phonetic differentiation into A, B, and C; 
although, of course, the notions /? and y may have existed previously. The 
parallelism of sound and meaning cannot emerge until both sounds and meanings 
have become distinct. There is often a period of unsettled usage previous to 
the definite association of each form with a meaning of its own (Wundt II 462 
sqq.). Instances of this type are of interest to the present study only when, 
after the differentiation, one of the forms can be shown to have influenced the 
other. It is necessary to discuss the various types. (Cf. Weekley, Words 139 
sqq. on doublets). 

It might be remarked, first, that the priority of the phonetic differentiation is 
not absolute. In the English words of and off, the distinction between the 
prepositional and the adverbial meaning existed before the phonetic differentia- 
tion, which is, on the contrary, a consequence of tne semantic difference, since 
of is the unstressed prepositional form, and off the stressed adverbial form. 
There are many instances of this kind (cf. Fowler, MEU, s. v. Differentiation). 

Delbriick (Grundfragen 154 sqq.) denies the existence of a correlative change 
in Wundt's sense, and he is no doubt right in one way: there is no direct symbolic 
value in the sounds of words, as Wundt attempts to vindicate for Rabe-Rappe 
(Wundt II 476). But there are more than the two types mentioned by Delbriick. 
I follow A. Erdmann (Dubbelformer) in his classification, and I refer to his lists 
for additional instances (see also Greenough & K. 345 sqq., and cf. Stocklein's 
remarks 75 sqq.) 

Many of Wundt's instances are of the type chose-cause, frele-fragile, raide- 
rigide (I.e. II 463), that is to say, one of the forms represents an unbroken 
development from Latin to French, the other is a loanword. As far as I can see, 
the development of each of these forms is in no way remarkable. If the forms 
happen to be similar, and their meanings also resemble each other, reciprocal 
influence is of course possible, but as I have shown in the previous section, the 
likelihood that synonyms influence each others' meanings is not great. English 
instances of this type, in which the development of each form takes place in isola- 
tion from the other, and leads to more or less different results, are chance and 
cadence; corps and corpse, which have now no sense in common; copula and 
couple, in which pair couple has been used sometimes to render one of the technical 
meanings of copula, but which are otherwise separated in sense; choir [quire) 
and chorus, on the other hand, are partial synonyms, the reason possibly being, 
as in couple, the employment of the English form to render the Latin. Hale and 
whole are a parallel instance, in which the Northern form hale has been taken up 
into standard English in a sense that is foreign to the normal form whole. 

In none of these cases have I been able to find conclusive proof of semantic 


influence one way or the other, except in the manner indicated for couple and 
choir, that is to say, an intentional substitution of one form for the other. 

If doublets are due to the levelling out of different forms in the paradigm, as 
for instance belly and bellows, hole and hollow, F. plier and ployer, they are at first 
total or almost total synonyms, but the difference of form facilitates different 
semantic development. Thus, hole has all the senses of holl [howe) sb. and 
hollow, sb., with a fuller development of its own. Shade and sAadow have many 
different senses; a third variant form, shed, is assumed by the NED to be influenced 
by the synonymous shud, and shows a distinctive semantic development. Morn 
and morrow, harry and harrow, are also differentiated in meaning. 

If the doublets are due to different stress of one word, as for instance of and 
off, through and thorough, it is evident that the semantic differentiation is prius, 
since it is the cause of the phonetic differentiation through strong and weak stress. 
There is no correlation, in Wundt's sense, until the phonetic process has set in. 
It would need a very thorough investigation to decide if a certain sense is due 
to independent development or to influence from the correlative word. In a, 
an, my, mine, and other pairs, there is a phonetic differentiation, due to different 
position, without any differentiation of basic meaning. 

Engl, then and than represent still another type. Forms with -e- and -a- were 
used promiscuously as adverb and conjunction in OE and ME. "When the 
adverb was reduced to J>en, from the 15th c. spelt then (sc. instead of the fuller 
form henne) there was a strong tendency to spell the conjunction in the same 
way; but in the 17th c. the tide turned, and by 1700 or a little later the con- 
junction was differentiated from the adv. as than. As the latter was, and is, 
pronounced [d^n'], it is manifest that it might be written either then or than with 
equal approximation to the actual sound" (NED). It is reasonable to assume 
that such a differentiation of form and meaning is due to printing practice, to 
the influence of leading writers, or to the influence of standing phrases in which 
one or the other form is traditionally used. There is no proof that the two 
words have influenced each other's meaning. 

We have, finally, pairs of doublets in which one word is the regularly developed 
form, the other an analogical formation, for instance last and latest. The con- 
nection with the positive late is more strongly felt for latest, which is no doubt 
the reason why latest refers to time, and last to order of succession. Latest is an 
instance of morphological combinative analogy. Brothers is the usual plural 
of brother, formed analogically, while the old form brethren has been retained in 
some transferred senses, probably because of its use in religious phraseology. 
A Swed. parallel is klddning and Manning, from the verb kldda 'to dress'. The 
latter represents the normal development of an older klddning, with assimilation 
of -dn- to -nn-, and keeps the concrete meaning 'dress'. The former is a new 
formation on the analogy of other participial nomina actionis, and has the 
abstract sense of 'dressing', owing to its closer association with the parent 
verb kldda (Noreen, Orddubletter iii). Germ. Reiter and Riiter (Wundt II 464 
and 476) belong here. 

In the last type we thus have combinative analogy (morphological) leading to 

230 GUSTAF STERN 9.35 

the creation of doublets. In the other types there is no trace of analogy, if we 
do not want to call by that name the levelling out in the paradigm represented 
by belly-bellows, etc. 

With regard to the principal question, that is to say, the question as to whether 
words belonging to such pairs influence each others' meanings there is no 
reliable evidence. As far as I know, the history of no such pair has as yet 
been investigated in detail. On the whole, I do not think that influence of this 
kind is likely to have been effective on a large scale. When two words have 
once separated, as harry and harrow, they are partial synonyms, and I have 
already shown that reciprocal semantic influence between synonyms is probably 
not strong. Even if they are felt as belonging to the same derivational group, 
that is not sufficient cause for analogical influence to be effective. But it is not 
possible to state anything with certainty as long as the actual historical facts of 
the case are not on the table. 

9.4. Phonetic Associative Interference. 

9.41. Preliminary Remarks. Phonetic associative interference 
differs in several ways from the two types of analogy previously dis- 
cussed. Combinative and correlative analogy were conditioned pri- 
marily by semantic similarity, while phonetic resemblance, where 
it occurred at all, played an inferior part. Associative interference 
is based primarily on phonetic similarity, although it seems that some 
semantic resemblance, or at least connection, is necessary. Combina- 
tive and correlative analogy consist in the filling out of an alread}*^ 
existing semantic group, and the existence of such groups is a condition 
of the process. Associative interference affects words that are not, 
or are not felt to be, members of any group, and on the basis of phonetic 
resemblance connects them with a better known word, or group, so as 
to give them a certain support for the linguistic feeling of the speaker. 
It thus leads to the formation of groups that are not historically 

Combinative and correlative analogy are of the type (described in 
7.34) in which the change is effected by the speaker, who uses a word 
in a new way. The hearer's part is to understand the novel use of the 
word. In associative interference, as in permutations and adequations 
(cf. 7.34), the change is the result of an association that offers itself 
both to speaker and hearer in the use of a word or phrase, an associa- 
tion that may be accepted if found serviceable (cf. Shand 70, on the 
laws of association and organization). 

It might be asked if, when there are so many and so essential differ- 


ences between associative interference and the two other types of 
analogy, we are justified in taking them together in one class. In so 
doing, I think I am in agreement with the usual practice among 
philologists, and, even if the analogy is different in nature and rests 
on a different basis, it is still a kind of analogy. 

9.42. Phonetic Resemblance as a Cause of Association. Since 
associative interference consists in group formation on the basis of 
phonetic resemblance, it is necessary to investigate the part played 
in associations by the phonetic elements of words, as compared with 
the semantic elements, which play the most important part in the 
combinative and correlative forms of analogy. Various experiments 
have been carried out, which show that there are two main reasons 
for phonetic associations {Klangreaktionen) . One of them is deflected 
attention; in ordinary circumstances it is the meaning of the stimulus 
words that determines the response (cf. 9.13), but when attention is 
deflected owing to fatigue, carelessness, hurry, etc., similarities of 
sound are apt to take the place of meaning as determining the reaction 
(Cf. Jung-Riklin IV 25, with a convenient summing up of the position. 
Dauber 197 sqq., Menzerath, Rignano 349, Claparede 243 — 244). 
Phonetic reactions in the shape of repetition {Nachsfrechen) are well- 
known in infant speech (W. Stern, Psych. 130). It is not probable 
that this factor is sufficiently common in the speech of adults to 
influence linguistic development to any appreciable degree. 

The second main reason for phonetic associations is the lack of known 
meaning in the stimulus word. When Dauber used nonsense syllables 
as stimuli, phonetic similarity was absent only in 3 — 5 % of the res- 
ponses, while with meaningful words such similarity was absent in 
nearly 77 %; it is of course not altogether to be avoided owing to the 
limited number of sounds in the language. Most of Dauber's observers 
(12) preferred rhyming responses; four of them preferred alliterative 
responses, and eight responded with words having the same suffix as 
the stimulus. The tendency to identify nonsense syllables with some 
known word may be so strong as to make useless all attempts to induce 
a neutral attitude on the part of the observer, or to get a purely 
mechanical response from him (Helson 37, 47. Cf. Stern, Kindersprache 


Whether phonetic resemblances play a part in combinative and cor- 
relative analog\' is as yet uncertain. Thumb's investigations (G. R. M. 

232 GUSTAF STERN 9.42 

III) have given a negative result, as also those of Eberschweiler (Zs. f . 
Psychiatrie 65, 240 sqq., esp. 246 — 247). It is only in the case of foreign 
or otherwise unknown words that the form seems to play a more impor- 
tant part. That the tendency to phonetic associations is strong in the 
case of unknown or incomprehensible words has also been demonstrated 
by Wartensleben (Zs. f. Psych. 57) and Peters (ib. 56). Meaningful 
responses to nonsense syllables are generally determined by sound. 

Esper (Exp. Inv. 46) concludes that tendencies towards associative 
interference in the form of contamination were strongest when the 
linguistic material deviated from English habits of syllable division. 
Bally (I 34) has pointed out that in learning a foreign language we are 
liable to attend more to the form than to the meaning of the words, 
because the latter requires a greater effort. The better the meaning 
of a word is known, the more do the associations attaching to the 
meaning tend to exclude those that attach themselves to its form, and, 
conversely, the lack of associations with the meaning of a word leaves 
the field open for associations of form or even of sound. 

Hey (Alyly 13, 203) states that words belonging to the present type 
are mostly lacking in linguistic or in objective support. The former are 
words not belonging to any semantic group (basic group); the latter are 
words lacking an easily recognized and in itself clearl)^ defined referent, 
as tree, house, sea, stand, sleep, long, black, over, left, etc. Archaic 
words are therefore especially exposed to associative interference, as 
well as form-words, words for relations, articles and pronouns. In my 
opinion, a strong linguistic tradition, founded on frequent use, is the 
best safe-guard for a word, while rare isolated words are liable to be 
influenced even if they denote referents of the kind mentioned by Hey 
(cf. Paul, Prinz. iii, 217 sqq.) 

The fact that homonyms exist in large numbers without any tendency 
to confusion is a warning against hasty assumptions of phonetic a.ssocia- 
tion. I may quote the case of the two English verbs to let, OE letan 
and lettan, the latter meaning 'to prevent'. The}^ partly fell together in 
form in ME, and if formal co-incidence were sufficient to cause mutual 
semantic influence, the circumstances were highly favourable. But the 
only point where the NED assumes such an influence to have taken 
place is where the meanings of the two words approach each other, 
from different starting-points, and by a development which might just 
as well have occurred in each even if the other verb had not existed. 


Let from letan signified, amcmg other things, 'to leave undone, to omit 
or forbear to do something, to desist, forbear'. Let from lettan had the 
senses 'to check or withhold oneself, to desist, refrain'. Here the two 
verbs were naturally confused, but the confusion does not seem to 
have touched their other senses. In the same way, the mutual influence 
of hless and hliss rests on similarity of meaning (see KED, s. v.). 

Phonetic interference is to be assumed when a word for some reason 
— foreign origin, dialectal origin, rare occurrence — is strange to speak- 
ers, which makes them involuntarily attempt to associate it with some 
known word or word-group, in order to procure a support for its mean- 
ing. But if the referent and meaning of a word are well-known, owing 
to sufficient frequency, so that it is well established in the memory of 
speakers, it can very well exist without the support of a group. 

To this type belongs so-called poptdar etymology, which is also one 
of the most popular subjects of writers of popular books on philological 
questions. I refer for literature and instances to the current handbooks 
and to the following works: Paul, Prinz. 218 sqq, B. I. Wheeler, Nyrop 
IV 322 sqq., Jespersen, Gr. a. Str. 69, Falk, Bet. 103, Noreen, Spr. 
Studier I 38, Sunden, Ell. Words 144; Ostberg 75 sqq.^) Stern, Kinder- 
sprache 417 sqq. gives lists of childish etymologies. 

9.43. Change of Meaning, hut not of Referent. Various classifica- 
tions have been proposed for sense-changes of the present class (Falk, 
Betyd. 97 sqq., Hatzfeld, Bed.-versch., Hey, ALL 13, Kjederqvist, 
who also gives a historical sketch)^), but most of them include types 
which have already been placed in the previous sections; many of the 
instances classed, moreover, do not involve any appreciable sense- 

^) The following is a curious instance: E. redshire, redshare or redshort is an 
adaptation of Swed. rddskdr{t, from Swed. rod 'red' and skdr{t 'brittle', and 
denotes iron that is brittle while in a red-hot condition owing to excess of 
sulphur in the metal. Similarly coldshort 'brittle in its cold state', of which the 
NED says: "This, and the parallel redshort . . . point by their early forms to 
adoption c. 1600 from Scandinavian, and prob. from Swedish, metallurgical 
terminology". If the words are associated by linguistic feeling with the 
common English adjective short (of. NED s. v. 20 and 21), this seems to be a 
folk-etymology, which would belong to the present type. 

*) Both the types described by Wundt (I 474 sqq.), Wortassimilationen mit 
hegrifflichen Nebenwirkungen, and Wortassimilationen mit Begriffsumwandlungen, 
belong here, the difference between them being one of degree only. Wundt 
has no instance of my other type, in which the referent also is changed. 

234 GUSTAF STERN 9.43 

change. I prefer to make a fresh distinction, between associative inter- 
ference leading to a change of meaning, but no change of referent, and 
associative interference leading also to a change of referent. 

In the first type, the association, on a phonetic basis, of the word 
with another word leads to a modification of the manner in which the 
referent is apprehended, but the referent itself remains in fact the same. 

Sandhlind is probably a perversion of OE sam-blind 'half-blind, dim- 
sighted, purblind'. The first syllable, at a later period, became incom- 
prehensible, and was interpreted as sand-. Cf. Johnson's explanation 
in his Dictionary: "having a defect in the eyes, by which small particles 
appear to fly before them". The noun standard has been affected, in 
many of its senses, by an association with the verb to stand; it is possible 
that in some uses it should be regarded as an alteration of stander. 
Brothel originally signified 'a worthless abandoned fellow, wretch, 
scoundrel; an abandoned woman, a prostitute'. In the combination 
brothel-house it was confused with the synonymous bordel, hordel-house, 
so that the simplex broHiel was apprehended as identical with hordel, 
and so received the meaning 'house of ill fame'. 

Bully 'tyrannical coward who makes himself a terror to the weak' 
is in the popular etymological consciousness now perhaps associated 
with bull, sb. (NED). Note that the verb bullock occurs in the sense 
of 'to bully'. E. welcome is from OE wilcuma, of wil or will 'pleasure, 
desire', and cuma sb. 'comer, guest'. Subsequently the first element 
was changed to wel- well-, and the second identified with the imperative 
or infinitive of the verb come, under the influence of OF bien venu, bien 
veigniez, I,, bene venisti, bene venies, and possibly also of Scandinavian 
forms. As far as can be gathered from the NED, this change may 
have been instrumental in abolishing the nominal sense 'one whose 
coming is pleasing', and possibly also in developing the vocative use as 
a form of address. Andiron is from F. andier, Modern F. landier, "In 
Eng. the termination was at an early date identified with the word yre, 
yren, iron, whence the later illusive spellings and-iron, hand-iron. In- 
stances also occur of land-iron, after later Fr." (NED). 

Wundt quotes German Friedhof, which is now, to German linguistic 
feeling, definitely associated with Friede 'peace'. The word is in MHG 
vrithof 'an enclosed court'. Similarly Siindflut 'the Deluge' is associ- 
ated with SUnde 'sin', but the original form is sin-vluot 'a general flood- 
ing' (Wundt I 480, with numerous instances). F. courte-pointe is from 


OF coutc-pointe, of coutte, a variant of couette 'covering', and pointe, 
past participle of the verb poindre 'piquer'. It has been apprehended 
as being a compound of courte 'short' and the noun pointe. OF 
soufraite 'privation', ly. suffracta, horn sub-frangere, has given souffreteux, 
which has been associated with souffrir. Lais is verbal noun from 
laisser, but is now associated with leguer 'to leave by will'; it is written 
legs, and sometimes even pronounced so. (Saussure 119, 238, with 
numerous instances). 

As thin as a wafer: the noun originally signified 'thin crisp cake', but 
the expression is now commonly associated with the sense 'a small 
disk of flour mixed with gum and non-poisonous colouring matter, or 
of gelatine or the like similarly coloured, which when moistened is used 
for sealing letters, attaching papers, or receiving the impress of a seal'. 

ME start 'tail' was used in the phrase start naked. When start became 
obsolete (the last quotation in the NED is from 1300), s^ar^ was substi- 
tuted. We find stark used as an intensive in ME: stark blind 1387, 
stark dead 1375. It is therefore not impossible for stark naked to have 
originated without any association with an earlier start naked.^) 

9.44. Change of Both Meaning and Referent. Belfry originally has 
nothing to do with bells, but denoted a 'tower used in attacking fortress- 
es'. Owing to the association of the first syllable with the word bell, 
belfry is now used almost exclusively to denote a tower containing bells. 
Shamefaced is a remodelling of shamefast 'modest, bashful', and the 
association with face has changed meaning and referent, so that the 
word now means 'showing shame in one's face'. Dispiteous 'unpitying' 
is a revival or contamination of earlier despite-ous 'full of despite' (see 

1) Jespersen (Language 172) thinks that the use of stark naked for start naked, 
as well as mate 'husband or wife', for make, milt 'soft roe of fishes' for milk, bat 
'vespertilio' for bak, trunk 'proboscis of an elephant' for trump, and others, are 
due to confusion in the minds of children. As in the case of stark for start, the 
displaced word seems to be without supporting associations in the relevant 
meaning, and this fact is, I think, a sufficient explanation for the substitution, 
Avithout having recourse to childish ignorance. Very few grown-up people have 
any knowledge of etymology and foreign languages. I regard all these instances 
as ordinary cases of popular etymology. Childish corruptions are, of course, 
numerous, but they cannot be expected to find their way into the standard 
language since they are laughed at and corrected by the grown-ups. Jespersen 
gives instances of this kind (1. c. 122). It will be evident that they are of another 
type. Cf. 7.42. 

236 GUSTAF STERN 9.44 

Fowler, MEU s. v. Revivals) . To curry favour, originally to curry Favel, 
F. estriller or torcher Fauvel 'to curry the fallow horse', from Roman de 
Fauvel (1310), the hero of which is a counterpart of Reynard the Fox. 
It is not clear whether before the date of this poem a 'fallow' horse was 
proverbial as the symbol of dishonesty. The phrase originally meant 
'to use insincere flattery'. Later, it was transformed to curry favour 'to 
seek to win favour, or ingratiate oneself with another' (NED. Jespersen, 
Language 173). Watershed is originally 'the line separating the waters 
flowing into different rivers or river basins; a narrow elevated tract of 
ground between two drainage areas, = water-parting'. It is also em- 
ployed "loosely" (ISTED) for 'the slope down which the water flows from 
a water-parting', or for 'the whole gathering ground of a river system'. 
The original meaning is perhaps from German Wasserscheide, which 
became common as- a scientific term about 1800. The earliest instances 
of watershed (in this sense) are from 1803. The new meaning is probably 
due to association with the verb to shed: the watershed is the area in 
which the water that flows into the river is shed or dropped. Cf. Fow- 
ler, MEU s. V. 

To pester is at first 'to clog, entangle, embarrass' etc., and is appar- 
ently an aphetic form of empester, impester, although these are only 
found later than the simplex. It was influenced in iis meaning by pest, 
and so came to mean 'to annoy, trouble, plague' (Weekley, Words 167). 

"In the seventeenth century, ingenuity had still its proper meaning 
of 'ingenuousness' or candour. Locke, for instance, could speak of an 
opponent's mode of argument as "more creditable to his acuteness than 
to his ingenuity", which to modern ears sounds like a distinction without 
a difference. But long before Locke's time the adjectives ingenious and 
ingenuous had become confused in popular use; even some very learned 
writers (or at least their printers for them) occasionally fell into the 
mistake of substituting the one for the other. Hence the noun ingenuity 
was often ignorantly or carelessly misused for 'ingeniousness' or 'ingeni- 
osity', and as these latter are both awkward words, while a noun 
answering to ingenious was more frequently wanted than one answering 
to ingenuous, the wrong sense ended by expelling the right one from 
the language" (Bradley, Making of Engl. 206). 



10.1. Theoretical Discussion. 

10.11. Preliminary Remarks. It is only in a minority of cases that 
shortening leads to a sense-change. Shortenings that do not lead to a 
sense-change are of inferior interest to the present study, but it is 
necessary to include them in the theoretical discussion. 

The problem of elliptical or brachylogical expressions has been the 
subject of much controversy. I shall have occasion in the course of 
the present chapter to mention some variant opinions, but many dis- 
puted questions may be passed over in silence, since they do not con- 
cern the problem of sense-change and thus fall outside the scope of 
my investigation. 

It is a regrettable fact that we are more handicapped by lack of 
material with regard to shortenings than in any other of the classes 
treated here, perhaps with the exception of adequations. For all that 
has been written on ellipsis, it has occurred to no one to make a histor- 
ical study of some representative group of elliptical or shortened 
expressions of the kind that involve a sense-change.^) The matter is 
of some importance, since it is often doubtful whether we should ascribe 
a change to shortening or to some other cause. 

The shortening in question is primarily a morphological process, the 
dropping out of some significant element or elements of the verbal 
form in the course of linguistic development. When the term is inter- 
preted in this way, we can speak of shortening only with regard to 
expressions that have a history. This is the case, in the first place, with 
single words; secondly, with what I shall call habitual combinations, 

^) For literature on the subject, see especially Wellander, Studien II 96 sqq., 
and Horn, Sprachkorper 124 sqq. 

238 GUSTAF STERN 10. 1 1 

in contradistinction to occasional combinations}) Instances will be 
given below. 

Historical shortening may occur in occasional combinations in certain 
circumstances; an occasional combination may be constructed on a 
sentence-scheme which can be shown to originate from an earlier, more 
complete scheme (see 10.14). Such a combination cannot be traced 
back to an individual prototype, but the scheme on which it is built 
could not have become what it is except through the omission of some 
element or elements. Such combinations are, for instance: he went to 
his uncle's, it costs five and six. The omitted words are not always the 
same, but vary from instance to instance. Sunden (HHipsbegr. 217) 
calls this typical ellipsis, because it is the type that is shortened, and 
distinguishes it from individual ellipsis, in which it is a habitual com- 
bination that undergoes shortening (see further 10.14). 

As explained above (4.27, 5.51 — 5.53, 6.32) no expression is ever 
really "complete" (cf. Paul, Prinz. 313 — 314). Completeness of ex- 
pression is a matter of degree; some expressions are definitely of normal 
completeness; others quite as definitely shorter than normal. The 
first task for the theoretical discussion is to find a criterion for deter- 
mining what expressions should be reckoned as incomplete or shortened- 

Four possibilities present themselves. We may compare the expression, 
as it stands, (i) with its referent, or (2) with the mental content that 
serves as its meaning; or (3) with other, more complete expressions for 
identical or similar mental content, or (4) with its own form at an 
earlier stage of development. Each of these possibilities will have to 
be discussed. On essential points, I make use of the arguments 
adduced by Sunden in his essay Om ellipshegreppet, to which I refer. 
It is necessary to state at once that I shall distinguish two main, 
types of shortening, clipping, or the shortening of a word; and 
omission, or the total dropping of one or more words. The terms 
will be more precisely defined below. 

10.12. The Referent as the Criterion of Completeness. It might 
be possible for an expression to be considered incomplete in relation 
to the referent which it serves to denote. 

With regard, first, to clippings, or shortenings of single verbal sym- 

^) The distinction corresponds more or less to Wellander's distinction between. 
semasiologisch einheitliche Verbindungen and semasiologisch lockere Gefuge (Stu- 
dien II 97). 


bols, we have to remember that a symbol — with the exception of 
echoic words — does not picture or mirror its referent, and there is 
thus no sense in speaking of the shortness of a symbol in relation to 
its referent. Simple symbols, e. g., war and peace, may symboUze 
referents of immense complexity. (See 4.27 and 5.51). 

Turning, secondly, to the total omission of words, the relation be- 
tween what is actually expressed, on the one hand, and the referent, 
on the other hand, cannot give any reliable guidance in distingmshing 
complete and incomplete expressions. No phrase is ever a really "com- 
plete" symbol of the referent that it serves to denote, in the sense that 
each item, or at least each important item, of the referent is symbolized 
by a separate verbal element. The speaker always leaves much unsaid, 
and the hearer has always to supplement what he actually hears with 
numerous items from the context, including signals. 

Speech is above all a practical means to an end, and to make it too 
elaborate would be to defeat its own purpose. The speaker's interest 
in the elaboration of the verbal form ceases at a point beyond which 
the effort would absorb an unreasonable amount of the mental energy, 
encroaching on the attention given to the topic of speech, without 
in any way improving the effect. An artistic drawing in which the 
salient features of an object are alone represented, may give a better 
idea of the object than a photograph that reproduces every detail. 
Mutatis mutandis, the same holds true of speech (Cassirer, Phil. Ill 44). 
On n'a pas besoin d'exprimer dans le language ce qui ressort necessaire- 
ment des circonstances (Nyrop IV 59. Cf. Biihler, Krise 40, Paul, 
Prinz. 313 — 314). Incompleteness of speech as compared with its 
referent is the normal thing; completeness is never attained and is not 
desirable. The incompleteness in this respect is a matter of degree; 
there is an unbroken series from a normally full expression to a lift of 
the eyebrows, "speaking volumes". At no point is there any possibility 
of drawing a definite line that might separate complete from incomplete 

10.13. The Mental Content as a Criterion of Completeness. Since 
the mental content of any expression is the user's apprehension of 
the referents, it is evident that what has been said in the previous 
paragraph concerning the inadequacy of the referent as a criterion of 
completeness must apply also to the mental content. We never express 
in words every detail of the mental content that we want to convey. 

240 GUSTAF STERN 10. I3 

However, since theorists have hitherto mainly worked with word 
and meaning only, there have been many attempts to prove that 
shortness or ellipsis is equivalent to an incomplete rendering of the 
mental content, or, rather, in some instances at least, of the supposed 
mental content. Such theories are found even in modern works, and 
it is therefore necessary to give a brief review of some relevant questions. 

The two main points of view may be called the logical and the func- 
tional points of view. 

(i). The Logical Point of View. Expressions have sometimes been 
called incomplete because they did not contain all the words that were 
thought to be required for a logically complete rendering of the mental 

The fundamental weakness of this criterion is the fact that there is 
no norm of logical completeness (cf. Sunden, Ellipsbegreppet 213 sqq.-). 
Many items which, on a formal logical analysis, are necessary links in 
the chain of thought, are habitually left undenoted; nevertheless, the 
expression is not felt to be incomplete in any way; on the contrary, a 
logically complete expression would often be unusual, or even abnormal. 
Hearers are accustomed to the shorter type, but not to the longer one. 
We do not miss a sign for the relation between the two elements in 
horse-shoe or snow-hoots, because we know enough of these words and 
their referents to interpret the complex symbols correctly (cf. Pard, 
Prinz. 320). Nor do we miss a link in an instance like the following: 
/ shall be at A to-morrow , so I shall not he ahle to meet you, where the 
impossibility of being in two places at once is a logically indispensable 
premise for the conclusion. There is no reason for assuming that language 
was formerly more strict on this point (cf. lycrch, Satzglieder) . 

Phrases of the type would you like red or white roses? have been 
termed elliptical on the ground that a logically complete phrase would 
run would you like red roses or white roses? It seems to me of very little 
value for linguistic purposes to apply such points of view (cf. Sunden, 
Ellipsbegr. 215). Speech is a practical means to an end and should be 
judged by its own laws only. 

It might be argued that ivhite roses and red roses denote two different 
referents, that there is a factual duality of referents, and that we must 
assume speaker and hearer to think of, or refer to, roses twice; con- 
sequently there is a factual psychic duality, and the logical claim is 
justified. But note that the sentence in question does not say white 


roses or red roses; it proposes, primarily, a choice between two colours. 
Roses, the word we are discussing, comprises both the red and the white 
kind. The adjectives particularize the two colours between which 
the hearer is invited to choose; the noun denotes the genus to which 
the two species belong. The assumption of two distinct referents for 
roses seems to be a mistake, introduced by the reformulation. I prefer 
to follow Paul in classing such phrases as cases of einmalige Setzimg 
mit zweifacher Beziehung (Prinz. 314). That is, primarily, a syntactical 
analysis, but it seems applicable also from the psychological point of 
view, and with regard to the actual referents. 

From the historical point of view which I have adopted (cf. 10.15), 
we have to regard the sentence scheme on which phrases like red and 
white roses are formed, as equally primitive and normal as more complete 
schemes; they are alternative methods of formulating mental content 
of a certain kind. One of these schemes is undeniably shorter than the 
other; it is also more common. If we like, we may call such phrases 
short; they should not be called shortened. If we like, we may also term 
them elliptical; it all depends on the definition of that term. If we 
do, a strict distinction must be established within the category of 
elliptical expressions, between ellipses that show historical shortening, 
and ellipses that represent shorter alternative schemes. 

The logical criterion is not applicable to clippings, and thus fails 
in one important respect. 

There are also one-word sentences, as Fire!, which are no doubt 
instances of what Wellander calls urspriingliche Kiirze (see 10.5), and 
which should not be called shortened. Brugmann (Verschiedenheiten 
16) quotes Guten Tag! and Du! (threatening), and thinks that it would 
be a mistake to call them elliptical. Intonation, situation, and gestures 
supply what is not expressed in words. The syntactical incompleteness 
is unimportant. The shorter expression is comprehended as it is in- 
tended, just as well as a more complete phrase. Brugmann suggests 
that we should distinguish Kurzsatz and Vollsatz. Logically, sentences 
of this type are of course unsatisfactory, but since there are all degrees 
between complete and incomplete, we can nowhere draw a line. 

(2) The Functional Point of View. An expression has sometimes been 
called incomplete because it was assumed that the hearer, in order 
to comprehend it, would have to supplement certain elements without 
which the expression would not be a sufficiently complete rendering of 

Goteh. Hogsk. Arsksr. XXXVIII: i. 


the mental content it was intended to convey. We may call this the 
functional point of view, since it evidently suggests that the phrase 
could not, without the supplementation, perform in a satisfactory 
manner the communicative function of speech. 

This theory covers, in part, the same ground as the logical theory, 
but is still more open to objections. First, according to the views 
adopted in the present work, the hearer always supplements that which 
is communicated, by mental material from his own knowledge and 
observation; it does not seem possible to draw anywhere a definite 
line which would separate from the great mass of supplementations, 
the cases where the supplementation is a consequence of incomplete- 
ness. Allusive expressions may be very short indeed, compared with 
what they are intended to convey; veni vidi vici is a case in point. 
But that phrase is not shortened or elliptical. (Additional instances, see 
Bain I 48 — 49). 

Secondly, the theory does not take into account the part pla3'ed by 
reference. Wellander, quoting a passage in which the brothers Grimm 
are spoken of, first by their full names, and later only by their first 
names, Jacob and Wilhelm, states, "demnach ist die Ellipse nicht nur 
eine Auslassung, sie begreift zugleich eine Erganzung mit ein: das 
gemeinsame died kommt bei partieller Ellipse nur in einer, bei totaler 
in keiner der korrespondierenden Verbindungen zum Ausdruck, wird 
aber trotzdem jedesmal mitgedacht}) Das im Satze zuriickbleibende 
died der Verbindung wird also semasiologisch in der Weise erganzt 
dass das andere died der Verbindung, obgleich unausgedriickt, mit- 
verstanden wird"^) (Wellander, Studien III i — 2). 

I am unable to comprehend what purpose would be served by sup- 
plementing mentall}' the surname Grimm or its meaning; both speaker 
and hearer know perfectly well what persons Jacob and Wilhelm refer 
to when these names are mentioned the second time, and no more is 
needed. Our mental processes are extremely averse to unnecessary 
detours, and are always seeking short cuts; when adequate reference is 
provided and the context expresses the speaker's intentions, the 
functions of speech are filled, and there is no reason for assuming a 
supplementation which would serve no useful purpose. The third method 
(3 a) described in 10.16 would probably be employed in a case like this. 

^) My italics. 


The comprehension of the shortened expression is based on verbal 
context (cf. 10.17). 

Thirdly, it might be objected to the present theory that it gives un- 
due prominence to the communicative function of speech. The sym- 
bolic, expressive, and purposive functions might claim, equally, to be 
taken into consideration, and that would evidently complicate matters 
too much. 

Fourthly, it is, after all, the speaker who produces whatever shorten- 
ings occur; their causes and criteria should not be sought in processes 
in the hearer's mind. 

The functional criterion is based on ignorance of the normal processes 
of comprehension, in that it assumes supplementation to take place 
only when a sentence is shorter than normal; but since supplementation 
always occurs, it obviously cannot furnish a criterion for shortenings 
(cf. Sunden, Ellipsbegr. 199 sqq. with further references). 

10.14. The Sentence-Scheme as a Criterion of Completeness. An 
expression might be considered incomplete when lacking one or more 
words usually employed in expressing mental content of a similar kind; 
in other words, if the expression is constructed on a more concentrated 
sentence-scheme than is commonly the case.^) 

This point of view looks more promising, since in not a few cases it is 
really the contrast between a phrase and the habitual sentence-schemes 
that makes us feel or suspect an omission. Telegraphic language is an 
instance in point. 

When we regard as shortened the phrases / am going to Smith's {St. 
Paul's, the barber's), it costs two and six, gib mir den Schwarzen, donnez- 
moi le noir, (in English a prop-word is used for the omitted noun in 
such cases), the reason is, obviously, syntactical. According to "nor- 
mal" syntax, which is equivalent to habitual sentence-schemes, or 
language-forms, we are accustomed to find, after a genitive, a word 
denoting that which is owned by the referent of the genitive; after an 
attributive word we expect a word denoting the possessor of the quality 
or characteristic denoted by the attribute. The words pronounced 
involve a relation to some other, correlative, word — but the correlative 

^) Sunden (Ellipsbegr. 213) takes this criterion together with the logical point 
of view, describing it as a comparison with "den morfologiska utst3-rsel en bety- 
delsekategori normaliter ager". 


word is omitted. The lack of an immediate correlative constitutes a 
criterion for typical ellipsis. 

On the other hand, it will not do to compare the following phrases in 
the same way: would you like white roses or red roses? and would you like 
white or red roses? It is true that the two phrases represent two degrees 
of completeness, but neither of them is syntactically incomplete, in the 
strict sense of that word. They are equally normal, and the first is 
probably less frequent than the second. We must apply here Paul's 
verdict: "misst man allemal den knapperen Ausdruck an dem daneben 
moglichen umstandlicheren, so kann man mit der Annahme von EHipsen 
fast ins Unbegrenzte gehen. . . es gilt diesen Masstab aufzugeben" (Paul, 
Prinz. 313). As a matter of principle, it should be noted that the 
existence of two synonymous and similar expressions, one of which is 
more complete than the other, does not necessarily prove that the 
shorter of the two developed out of the other one by the omission of 
some element. Assist me on deck is probably not a shortening of assist 
me in getting on deck (Krueger, Arch. 107, 356; Paul, Prinz. 319). Each 
case must be judged on its own merits. 

The syntactical criterion must be restricted to the dropping out of 
words or phrases which are immediate syntactical correlatives to one of 
the words or phrases retained in speech. The correlation involves the 
consequence that both fundaments of the relation must formerly have 
been pronounced. In other words, the present, short, sentence-scheme, 
is the result of a historical development, starting from a more complete 
scheme. We cannot trace this development for the individual phrases, 
but only for the scheme. 

The syntactical criterion does not cover clippings. 

10.15. Historical {Individual) Shortening. We have, finally, the 
fourth criterion of shortening, the comparison of an expression, as it 
stands, with its own earlier form. As already pointed out above (10. 11) 
such a comparison can be made only with regard to expressions that 
have a history of their own, (i) single words, and (2) habitual combina- 

A shortened expression, then, is one which has originated, through 
clipping or omission, from an earlier, more complete expression; or, as 
Sunden formulates it (Ell. Words 35; cf. Ellipsbegr. 217): "the first 
criterion of the elliptical phenomenon that we want to establish, is 
that it offers an historical abbreviation." This point of view provides 


an unmistakable criterion as soon as we know the history of the 
words we are investigating, and that is an indispensable condition for 
all historical linguistic research. It covers both clippings and 
omissions in habitual combinations.^) 

With regard to the interrelation of the syntactical and the historical 
points of view, it is evident that the former does not cover clippings, 
but that it does cover historical omissions. The strict historical 
point of view, on the other hand, does not cover typical shortening. 
The logical and functional points of view likewise often cover omissions, 
but they fail to provide the necessary limit between shortened and not 

10.16. The Psychic Process. With regard to the psychic process in 
the speaker's mind when producing a shortened expression, we can say 
little. There is, however, no reason for assuming the process to be 
essentially different from the "normal" production of speech (cf. ch. 5). 
We have to assume that the speaker is able to judge, more or less 
automatically, on the basis of context (which comprises not only verbal 
context, but also perceptual and mental context of all kinds) what 
items may be left undenoted without detriment to the adequacy of his 
speech, or should be left undenoted so as to fulfil the functions of 
speech even better than a more complete expression would have done. 
No doubt he is guided by traditional sentence schemes in selecting and 
using the words that will meet his purpose in the momentary situation, 
and his long training in his mother tongue will have provided him with 
the necessary automatic skill. The shortened expressions are apparently 
of a limited number of types, all of them familiar to both speaker and 

The hearer is similarly equipped. The process of comprehension may 
be briefly analysed as follows. If a word is clipped, there are three 
theoretical possibilities, (i) The hearer may, by association of conti- 
guity, re-instate the missing portion of the word, and then comprehend 

1) I may quote the following definitions: "Ellipse ist nicht die sprachliche 
Ausdrucksform, der etwas am Sinne, sondern der etwas an Form fehlt . . . es 
kommt nicht dafauf an, was heute der Redende denkt, sonder wie jene Form, 
so wie sie ist, hat werden konnen." Krueger, Arch. 107, 355. "Unter Ellipse 
verstehen wir nacli § 366.5 die Erscheinung, dass ein Bestandteil einer usuellen 
Ausdrucksweise, der zur Mitteilung des augenblicklichen Vorstellungsinhaltes 
entbehrlich ist, unausgesprochen gelassen wird." Brugmann, K.vgl. Gr. § 936. 
See also Bally, II 278 — 279. 


the whole. (2) The remaining portion of the word may be character- 
istic enough to indicate the referent without recall of the missing part, 
and without the help of context.^) Both these methods of interpretation 
are instances of the law of association which states that any element 
that has previously occurred as member of a complex, on being appre- 
hended alone bearing its membership character, tends to reproduce the 
complex. Such re-instatement may go either of the two ways just 
indicated, via the whole word to the referent and meaning, or directly 
to the referent and meaning; in the latter case we should perhaps also 
assume a subsequent re-instatement of the whole word; however, since 
it would serve no useful purpose it seems probable that in many cases 
no such supplementation of the missing part takes place. As far as I 
know, the matter has not been investigated in detail. (3) The hearer 
may know beforehand, or infer from the context, what referent the 
stump word or the pronounced part of the original combination is in- 
tended to denote; he can thus identify the shortened expression with 
the full expression, because it refers to the same thing. The frequency 
of this form should not be tmderrated. Suppose two persons to be 
talking and one of them to be using the word bus, till then unknown 
to the other. The two situations in which this is likely to happen most 
often are (a) that the interlocutors are speaking about omnibuses, and of 
course are aware that they are speaking about omnibuses; any new 
name applied to that referent will at once be understood by the hearer, 
because it refers to a thing he knows of; (b) that the hearer will be 
aware, by the speaker's gestures or looks, that the unknown word re- 
fers to that big vehicle which is approaching along the street; and since 
he sees what that is, he will know what a bus is. 

Since shortenings are very often produced during the course of a 
speech in which a more complete name has at first been used, the first 
of these forms of comprehension (a) is of especial importance, and I 
shall have occasion to recur to it more than once. The shortening is 
conditioned by mental or verbal context. In the second form (b) the 
shortening is conditioned by perceptual context (cf. 10.17). The 

^) As pointed out above (6.1 and 6.3) this often occurs even if the whole word 
is pronounced by the speaker: the hearer notices some characteristic and salient 
portion of it and that may be sufficient for him to interpret it correctly; he then 
pays no attention to the remainder. Cf. also Sund^n, EU. Words 52, with quota- 
tions from Passy, Behaghel, and Jespersen. 


correctness of the comprehension is at once verified by its fitness, or 
the reverse, to the situation, and the likeness of the stump word to the 
full name. 

Taking the matter theoretically — we cannot do otherwise as long as 
we have no special investigations on this point — the method of re- 
instatement seems more likely to be employed when the referent has not 
been previously mentioned, but is designated for the first time by the 
clipped word, and when it is not present to the perceptions, or otherwise 
contextually indicated. In all other cases the third method would 
probably be quicker and easier, and thus more likely to be adopted. 
However, considering the immense variability of the mental processes, 
we cannot say what may happen in individual cases. It follows that 
the process of comprehension is useless as a basis of classification, even 
if we disregard the fact that no such processes occur in the speaker's 

The same principles, with some modifications, appl)' to the compre- 
hension of phrases containing omissions. If the omission occurs in a 
habitual combination (see on this term 10.22) the method of re-instate- 
ment may come into use, but the third method, that of context, is just 
as likely to be employed here as in the case of clippings. When the 
omission occurs in an occasional combination, the contextual method is 
the only one possible; as before, the context may be of any kind. 

The theory embraced by many writers, and exemplified above (10. 
13.2) from Wellander, concerning so-called supplementation, rests on 
the assumption that the words are the only basis for comprehension and 
the mental processes involved therein. According to the theory adopted 
in the present work, a clipped symbol may adequately perform its 
function of referring to a certain referent, and the referent then becomes 
the basis of the further mental processes, leading to the actualization of 
relevant items of the hearer's knowledge with regard to it. Similarly 
in the case of omissions, only with the difference that the reference to 
the item originally denoted by the word now left out, must always be 
supplied by context, including inference. 

It would seem that, with regard to shortenings, the symbolic function 
is especially important. An adequate fulfilment of that function will 
generally imply that the communicative function is also duly filled: 
the hearer will know what referents are intended. No doubt the ex- 
pressive and purposive functions may sometimes require words to be 


used that would not be absolutely necessary for the symbolic and com- 
municative functions; but, on the other hand, the two former functions 
are in oral speech often filled by signals: gestures, facial expression, 
and intonation, which are independent of formal shortening. These 
matters are, as far as I know, not investigated. The fact that the ver- 
bal element left out has no absolutely indispensable symbolic function 
is one of the chief conditions of shortening. 

I wish to emphasize that the processes involved in producing and 
comprehending shortened expressions are essentially the same as the 
normal processes that have been described in chs. 5 and 6. There is 
only a varying degree of completeness of expression. 

10.17. The Conditions of Shortening. The problem of the psjxhic 
processes involved in the production and comprehension of shortened 
expressions is intimately bound up with the question of the conditions 
of shortening. 

Generally speaking, we may say that the use of shortened expressions 
presupposes a comparatively full context: verbal, perceptual, or mental 
(cf. Paul, Prinz. 313; and 6.32 sqq. above). When we first speak of 
Jacob and Wilhelm Grimm, immediately afterwards using the first 
names only, Jacob and Wilhelm, the shortening is possible owing to the 
verbal context. Perceptual or mental context may function with sim- 
ilar effect. When the waiter at a Berlin Bierstube asks a guest, "Helles 
oder Dunkles?" the hearer may understand the question because it is 
what he expects to be asked in that place, or because he sees the two 
tumblers offered by the waiter; the context is mental or perceptual 
respectively. Similarly when an English waiter whispers in your ear, 
during the course of a banquet. Thick or clear, sir? or Sweet or dry, sir? 
the former question is referred to soup, and the latter to the champagne, 
because the diner expects those things at that stage of the entertain- 
ment (mental context), and it is mostly unnecessary for him to glance 
behind him in order to make perceptual context supplement the mental 

The more speaker and hearer know about the topic beforehand, and 
the more they agree in their opinion about it, the less need is there of 
verbiage in order to arrive at a perfect understanding. The operating 
physician and his trained assistant may work together throughout a 
long and complicated operation with only a word or a gesture now and 
then: the assistant adapts his acts to those of the physician because he 


knows his business, i. e., because of his mental context (cf. Biihler, Krise 
37 sqq. on Kontaktverstehen) . In such circumstances, the vaguest 
terms, or even gestures or looks, may be sufficient. Written language, 
which has to forego the assistance of perceptual context, is often com- 
pelled to be more explicit than spoken language. 

We may turn the matter round and say that a condition of shortening 
is that the verbal element or elements left out should have no indispens- 
able function. It is evident that this happens when the function that 
this element should have fulfilled is covered by the help of context, 
and the more complete the context is, the less need is there of detailed 
verbal expression (cf. Paul, Prinz. 313).^) 

Another, and scarcely less important, condition for shortening is 
constituted by the traditional speech habits. The items picked out for 
naming are determined not only with regard to the actual context, and 
the requirements of the verbal functions, but also with regard to the 
available means of expression, the stock of words, and, especially, the 
stock of sentence schemes, as well as the general speech habits of the 
community or group (cf. 5.51). The hearer is, normally, familiar with 
the speaker's language system, and is thus equipped for the task of 
interpretation. An unusual degree of concentration will at once present 
difficulties to the hearer. 

To sum up: the conditions of normal shortening are, (i) the speech 
functions that should have been fulfilled by the omitted elements must 
be covered with the help of available context. (2) The shortening 
should conform to the speech habits of the group. 

Shortenings that exceed these limits will present difficulties of com- 
prehension, and are therefore less liable to occur in normal easy speech. 

As might be expected, we find that the conditions of shortening are 
based (i) on the methods adopted by the human mind in handling 
speech as an instrument for its purposes, and (2) on the consequent 
construction of the instrument itself. 

These conditions involve both speaker and hearer. The actual, for- 
mal shortening is of course effected by the speaker, who is the judge of 

1) Noreen distinguishes deictic ellipsis, where the supplementation is made on 
the basis of the situation (as [give me) a cup of coffee), and anaphoric ellipsis, 
where the supplementation is based on verbal context (as gold {coins) and silver 
coins). In the former instance, the context is mental or perceptual. I do not 
regard the second quotation as shortened. 


its possibility and expediency. But since a speech must fail in its pur- 
pose if it is not correctly comprehended by the hearer, we may say that 
the limits to shortening are set by the hearer, or, more strictly, by the 
speaker's opinion concerning the hearer's ability to grasp his meaning. 

The conditions stated apply both to clipping and omission, but most 
obviously to the latter. When a word is clipped, it is generally desirable 
that the remaining stump should be distinctive enough to lead on im- 
mediately to the referent. 

An intermediate type may be specially mentioned. In a binary 
combination consisting of head-word and qualifier it sometimes happens 
that the whole fuses into a practically unitary appellation, out of 
which one link may be allowed to drop without detrimental effects, 
because of the close association to the meaning. Thus it seems to me 
that the abbreviation of ville capitate into capitale presupposes a certain 
degree of habituation. The outstanding instances of this type are the 
French negations: the negative particle ne is dropped in connection 
with pas, point, personne, jamais, aucun, etc. The omission could not 
have taken place if the whole expression ne — pas, etc. had not first 
fused into a unit, and been apprehended by linguistic feeling as signi- 
fying 'not'. The dropping of ne is therefore a clipping rather than an 
omission (cf. 10.3). 

10.18. The Causes of Shortening. 

10.181. Preliminary Remarks. The causes of shortening must be 
carefully distinguished from its conditions. Since shortening is involved 
in the speaker's method of using speech, the causes of shortening lie 
altogether with the speaker, in contradistinction to the conditions, 
which, as just stated, lie partly with the hearer. It is only indirectly 
that the hearer may cause shortening, by the speaker using a short 
expression out of consideration for the hearer, or to affect the hearer 
in some way. 

The causes of shortening may be phonetic, graphic, functional, or 
economic, and each kind requires separate discussion. 

10.182. Phonetic Causes. A word may be shortened for purely 
phonetic reasons, as any historical grammar will show. Phonetic 
reasons have also been adduced to explain shortenings of greater ex- 


tent, which appear to go outside the action of sound-laws in a strict 
sense^) . 

W. Horn (Sprachkorper) asserts that if a syllable is without linguistic 
function — he is thinking especially of the symbolic function — it is 
liable to drop off, even if the loss of sounds is greater than could be 
accounted for by the action of known sound-laws. The loss of function 
thus directly causes the loss of sounds. Objections to this theory have 
been raised from various quarters. K. Luick (E. St. 58, 244; cf. also 
E. Ekwall, Anglia Beibl. 34, 131) thinks that the loss of function di- 
rectly occasions a loss of expiratory stress. This factor closely follows 
the gradations in importance of the syllable. "Andernsich jene Abstuf- 
ungen, so andert sich auch der Atemdruck, negativ wie positiv; letz- 
teres im Fall der Hervorhebung einzelner Begriffe durch emphatischen 
Akzent. Hier haben wir also einen unmittelbaren, automatisch wirken- 
den Zusammenhang zwischen Wortvorstellung und Artikulation" 
(Luick, 1. c). The weakening of a sound or sound-group introduces 
new phonetic conditions which may lead to total loss, but only if the 
group is of a type that is regularly dropped in such positions in the 
language. The immediate causes of the loss are thus phonetic; the 
conditions for their action may be brought about by functional factors 
(See also Karstien, St. u. Aufg. 399 sqq.). 

We should probably add that, if a syllable loses stress owing to loss 
of function, it can afterwards be dropped, not only for phonetic reasons, 
but also for economic reasons, in order to save effort in pronunciation 
(cf. 10.185). Ini many of the instances that are adduced in this section, 
phonetic, functional, and economic causes are so blended that it is 
not possible to disentangle them. Sometimes, the phonetic process is 
only a secondary cause. 

Speakers have a tendency to slur unstressed and unimportant syll- 
ables, and hearers often pay little attention to them, even when they 
are pronounced (cf. 6.1 and 6.3). The articulatory movements may 
be too faintly innervated, so that no actual sound is produced until 
the second or third syllable (Jespersen, Ph. of Gr. 310, Sunden, Ell. 

^) Sunden 's term for this kind of shortening is prosodic reduction, which Wellan- 
der (Studien II 30') renders prosodische Reduktion. Cf. Sunden, ElHpsbegr. 228, 
Jespersen, Negation 6 and Grammar II 8. 91, Slettengren passim, Poutsma I 
•999 sqq., Kjellman 3, Carnoy 240, Horn, Sprachkorper 10 sqq., Bergmann passim, 
X. Miiller passim, Fowler, MEU s. v. Curtailed Words. 


Words 65). Instances of such prosiopesis are 'cause, 'cept, 'sterrics 
'struction, 'tickler, for because, except, hysterics, instruction, particular, 
Slettengren (144) quoting a number of instances from Kipling's Just 
So Stories, points out that they are common in children's speech, and 
explains them as due to imperfect reproduction. Other English instan- 
ces are tawdry and tantony, from St. Audrey and St. Anthony (see NED, 
Horn, Sprachkorper 16 and Weekley, Words 34). 

In words of command, which have to be pronounced with a violent 
effort of the articulatory organs, concentrated generally on one syllable,, 
the other syllables may drop out, as in attention! > teniion! > shun!'^} 
Near! is used for no near! an order to the helmsman to come no closer 
to the wind (NED, and Horn, Neue Beobacht. 135). 

Euphemistic motives are no doubt active in the numerous shortenings 
of expletives (Brugmann, Abkiirz. 369 — 370), as well as in the corruption 
of such words as swelp me < so help me God, cripes < hy Christ's stripes 
(Jaeger, Engl. St. 60, 292. See also Sunden, Ell. Words 43, Jespersen, 
Gr. a. Str. 243, Weekley, Words 65). Sblood, snails, slije, and zounds,. 
may be phonetic shortenings of God's blood, nails, life, wounds; the 
main motive is here perhaps euphemistic (Sunden 1. c). 

Prosiopesis may occur also in combinations: [Do you) see? {Do you) 
remember that chap? [Will) that do? (I'm a)jraid not. {When you) come 
to think of it. {I shall) see you again this afternoon (Jespersen 1. c). 

It is probable that economic causes co-operate with phonetic causes 
in many abbreviations. Brugmann points out that we are especially 
careless with words that do not convey a communication, as the con- 
ventional formulae of address: Exlenz for Excellenz, 'ntag, 'nacht for 
Guten Tag, Gute Nacht, 'schamster for gehorsamster Diener; French 
msieu for monsieur, sple for s'il vous plait; English {good) morning,, 
[kju] for / thank you, please for if you please. All kinds of courtesy 
formulae are thus shortened (cf. Horn, Sprachkorper 18). I am not 
sure that the condition of shortening is the fact that these expressions 
do not convey a communication. As a matter of fact, they do. The 
most important condition is, I think, the circumstance that they 
are regularly used in identically recurring situations; after sufficient 
repetition, everbody knows what is intended, even if the words are 
more or less mutilated; Good morning! may degenerate into an inarti- 

^) 'E sez to me, "Shun!" an I shunted. Kipling, The Seven Seas, Back to the 
Army Again. 


culate grunting sound. We should perhaps assume the economic factor 
to be the main reason for such shortenings. 

Context in the form of an unambiguous situation probably also 
conditions many shortenings of words frequently used in group dialects; 
Oerman officers, but not, for instance, business men, are liable to say 
Rement for Regiment (Brugmann, Abkiirz. 369 sqq.). 

Haplology may be a phonetic cause of shortening. Sunden, quoting 
Jespersen (Progr. in Lang. 343), instances familiar and vulgar pronun- 
ciations like [laibri, fehri, litri, probli] for library, February, literary, 
j)robably. Brugmann (Abkiirz. 370) gives Lat. semodius for semimodius 
and praestigiae for praestrigiae, which he calls Dissimilationserschein- 
tmgen. He thinks that the speaker at the moment of uttering the 
word has his articulatory movements not yet properly prepared, so 
two identical or similar movements may fuse into one. The tendency 
to save energy may also be active. 

An incorrect analysis of words may lead to shortening. Sunden 
(Ell. Words 54) instances atomy for anatomy, in which the first syllable 
lias been apprehended as the indefinite article. Egma for enigma occurs 
in Shakespeare. Several instances of this type have become standard 
forms: cater (sb., later caterer) is due to the loan-word acatour being 
taken for a catour; O. Vx. V assise was analysed as la sise, and in English 
perhaps assize was analysed a size; accrew > crew and appeal > peal 
are not quite certain according to the NED. Hydropsy has become 
dropsy in ME by the stages f)e ydropesie (with stress on the second 
syllable) > th'idropesie > the dropesie. Fence is an aphetic form 
of defence. (Should we assume haplological shortening of the defence? 
Cf. Weekley, Words 62 sqq., 113, 116; Jespersen, Language 173 on met- 
analysis. Grammar II 5.61 on numerus metanalysis, and Subtraktions- 
-dannelser) . 

10.183. Graphic Causes. Graphic shortening, like phonetic short- 
ening, is closely bound up with functional and economic motives. 
The written (or printed) form of an expression may be shortened for 
greater ease and convenience in writing, or for the sake of greater 
legibility or clearness. If the shortened symbol comes to be read as 
written, instead of being pronounced with its full form, we get a per- 
manent shortening. Gent for gentleman, ad for advertisement, sov for 
sovereign, are no doubt due to abbreviations in written or printed form. 
■Sunden (Ell. Words 70) calls these instances orthographic abbreviation, 


and states that they are particularly common in English. Wellander 
(Studien II 43 sqq.) speaks of graphische Kiirzung and grapMsch 
hedingte Wortunterdruckung, the former being clipping and the latter 
omission for graphic causes. 

Among graphic shortenings are to be reckoned the innumerable 
abbreviated terms in the form of two or more letters, generally initial 
letters, that are so common in group dialects. Doctors speak of the 
and d. t. (tuberculosis and delirium tremens), business men of fob and 
fow (free on board, first open water), timber merchants of t. g. h. 
(tongued, grooved, beaded), cricket players of /. h. w., and so on. During 
the war a large number of military abbreviations found their way into 
the papers and so into the common speech (GHQ, BEF, etc.), and 
some of them were employed as common nouns [wanes, members of 
the W. A. A. C; Dora for D.O.R.A., the Defence of the Realm Act; 
cf. Swed. Sara for S. A. R. A., Stockholms Allmanna Restaurant- Aktie- 
bolag). A typical Swedish instance is siins, now pronounced as written, 
originally st.-ins., the official abbreviation of stationsinspektor. Other 
official abbreviations in general use in Swedish are rek. and ass. for 
rekommenderad forsdndelse and asstcrerad jorsdndelse (registered letter, 
and insured letter). (See Nyrop, Etudes 13 sqq., 41 sqq., Kjellman 
16 on abbreviation in advertisements, 17 on abreviation officielle, and 
76 sqq. on abreviation par initiales; Rodhe, Abkiirzungen) . 

According to the NED, miss for mistress may have been suggested 
b}^ the written abbreviations Mis. and M'« (the latter representing the 
spelling mistris) which were common in the 16— 17th c. (Cf. Sunden, 
Ell. Words 67; Horn, Sprachkorper 19 thinks that the shortening is 
due to the loss of function of the second syllable). It seems probable 
that English forms like Aead., exam., prep., lab. [academy, examination, 
preparation, laboratory), are due to graphic shortening. This would 
explain why the syllable carrying the main stress of the words is 
sometimes dropped (cf. Horn, 1. c. 17). 

In many instances it seems quite evident that the striving for ease 
and saving of trouble is the cause of the graphic shortening; thus in 
the use of initials. In other instances the striving for legibility and 
clearness is quite as evidently active, especially in the use of sign- 
boards, tables, and the like. 

Another type belonging here is the shortening of street names 
instanced by Wellander (Studien II 119): Ecke der Jiidenund Stralauer. 


It should have been pointed out that such abbreviations are no doubt 
due to official notices and signs. Nyrop (IV 60) states that it is the 
metropolitain that teaches Parisians to say Alesia, Rome, for rue 
d'Alesia, rue de Rome. Similarly, the tramway station at the corner 
of Parkgatan and Viktoriagatan in Gothenburg is called Park- 

French Hotel Ramhouillet, Chocolat Suchard, Style Louis XIV , are 
quoted by Wellander (Studien II 54) as the pattern for similar German 
formations: Hotel Wagner, die Angelegenheit Miiller, die These Deutsch- 
bein, die Akten Schmidt, Fort Hdseler, Division Falkenstein, etc.^) 

The dropping of particles in habitual combinations in German has 
given rise to a real sense-change. Wellander states that in Ndhe Zoolog- 
ischer Garten, Mitte See, Ziige Richtung Berlin, Anjang Oktober, Ende 
Juni, the words Ndhe, Mitte, Richtung, Anjang, and Ende, have acquired 
prepositional meaning. Constructions of this sort occur chiefly in 
official and business language, perhaps originally, and still mainly, in 
superscriptions, notices, directions on signposts, tables, and in other 
circumstances where tabular form is convenient, and also conduces 
to brevity and clearness. The sense-change of Mitte, etc., is a change 
mainly of relational meaning (syntactical function) and thus falls outside 
the scope of the present work. 

The same appears to be the case with some German instances where 
a preposition has been dropped. In Kraft has been shortened to krajt 
{in Kraft allein des Rings > Krafft des Gesetzes); von Wegen to wegen 
(originally: von ains klainen misstrawens wegen, then, with postposition 
of the noun: von wegen des Priors, and finally, with dropping of the 
preposition: wegen deines V alters). The literary origin of several of 
these shortenings may, according to Wellander, indicate that they 
are mainly graphic, but phonetic motives are not improbable; we 
have probably to reckon with various causes (Wellander, Studien II 
62, with further instances; Horn, Sprachkorper 96). 

^) Wellander does not state his views on the French forms. According to Krue- 
ger, Arch. 108, 113, and Bergmann 28 they are formed on the analogy of older 
Hdtel-Dieu, la jete St. Jean, which show an uninflected genitive (or dative). 
Modem fin courant, fin juillet, commencement aoitt may be an analogical extension 
of this type. Cf. Krueger, Arch. 107, 359, and Bergmann 30. 

In English, the proper name is generally placed before the other noun: Reming- 
ton Typewriter, Zeppelin Airship, Waverley Hotel, and so on. 


I would suggest that these developments might be explained like 
English a dune > adown > down; that is to say, the original combina- 
tion of preposition and noun fuses into a prepositional phrase, appre- 
hended as a single preposition, and the first, unstressed, syllable is 
dropped, probably partly from phonetic and partly from economic 
causes. (Further instances in Krueger and Bergmann). 

10.184. Functional Causes. The functional causes of shortening 
are perhaps the most important, but they are also much more dif- 
ficult to define and discriminate than the phonetic and graphic 
causes, with which, as already stated, they often co-operate. 

Starting with the communicative and symbolic functions, I have 
already pointed out above (10.12) that brevity may conduce to a 
better understanding, and too many words confuse the point at issue; 
the picking out of a few salient items may give a better idea of the 
topic than prolonged wallowing in details; the hearer may understand 
the shortened expression quicker and better. 

The expressive function is partly covered by signals, but a short- 
ened expression by itself ma\-, owing to its unusual and perhaps 
ungrammatical, form, reflect better the speaker's emotional state and 
make the hearer aware of it. Clippings are very often intended to make 
the words express sympathy or endearment towards the persons 
addressed; nursery speech abounds in nighties, tootsies, etc., and clip- 
pings of proper names, transforming them into pet names, are often 
due to a similar desire for emotive effects (cf. Sunden, Ell. Words 49, 
53). The numerous shortenings (clippings) in slang and cant often aim 
at a humourous effect. 

Conciseness and brevity increase vivacity, and thus also the effective- 
ness of speech; brevity is the soul of wit; the purposive function may 
consequently be better served by shortened phrases. 

Causal classes cross the formal classes; different functional types 
may lead to similar results, and vice versa. There are, however, a 
few types that stand out distinctly. The most important of them is 
euphemistic shortening. 

Euphemism is due to a state of mind on the part of the speaker, 
characterised by a desire to tone down or veil offensive, indecent, or 
otherwise unpleasant statements. It very often leads to metaphors 
(see II. 8), but not seldom to shortenings; the offending word is either 
corrupted or clipped, and thus to some degree disguised, or else it is 


omitted altogether. Wellander (Studien II 9 sqq.) gives a number of 
German instances: in gesegneten Umstdnden is shortened to in den 
Umstdnden; ini Galgen hdngen to hdngen. The numerous abbreviated 
invocations of gods and devils belong here. (French instances in 
Kjellman 3). 

The euphemistic tendency may be based on consideration for the 
hearer, or on traditional avoidance of certain words, owing to taboo, 
or to Mrs. Grundy; but consideration of the speaker's own feelings 
may have similar results. A well-bred speaker will not use coarse words, 
but either leave them out or employ a paraphrase; the coarse words do 
not belong to the level of style and vocabulary at which he habitually 
moves, and which he will keep even if the hearer should happen to be 
less fastidious. 

10.185. Economic Causes. Sunden (Ell. Words 49 and 53) speaks 
of the lex parsimoniae "that so frequently influences linguistic utterance". 
The desire to save effort may lead to the leaving out of verbal elements 
that are not indispensable. Brugmann (Abkiirz.) emphasizes the 
importance of the Bequemlichkeitstrieb. Similarly Wittmann (116). 
Paul (Prinz. 313) seems to ascribe ellipsis a little too exclusively to 
this factor: "Die sparsamere oder reichlichere Verwendung sprach- 
licher Mittel fiir den Ausdruck eines Gedankens hangt vom BedUrfnis 
ab. Es kann zwar nicht geleugnet werden, dass mit diesen Mitteln auch 
vielfach Luxus getrieben wird. Aber im Grossen und Ganzen geht 
doch ein gewisser haushalterischer Zug durch die Sprechtatigkeit. Es 
miissen sich iiberall Ausdrucksweisen herausbilden, die nur gerade 
so viel enthalten, als die Verstandlichkeit fiir den Horenden erfordert". 

I have already pointed out that the labour-saving tendency may 
work indirectly, leading to shortenings of the graphic or phonetic 
types. The question is whether the economic motive may be a direct 
cause of shortening. Sunden assumes that when fall is used for fall 
■of the leaf, the reason is to be sought in this motive, and the explana- 
tion is probably correct. The human mind is so constituted that it is 
less of an effort to indicate only the genus to which the referent be- 
longs, or to indicate the referent with a gesture — which may take 
the verbal form of a pronoun — than to give the precise name. The use 
of pronouns, or of generic words, to save mental effort, is especially 
characteristic of undeveloped minds, unintelligent or immature. The 
demands of the speech functions must set a limit to the economic 
Goteb. Hogsk. Arsskr. XXXVIII: i. 


tendency; on the other hand, since shortness may serve the functional 
needs better than a more complete expression, the two tendencies 
must sometimes co-operate. Purely phonetic processes may intervene 
as a third part. The combinations thus brought about in individual 
cases would no doubt repay a closer study, but as yet they are very 
little known. 

10.2. The Classification o£ Shortenings. 

10.21. The Principal Point of View. For the main distinction be- 
tween different types of shortening, we have to choose between their 
causes, their conditions, and their form. 

The distinction between phonetic, graphic, functional, and economic 
causes of shortening is evidently of great importance both theoretically 
and practically. Nevertheless, it cannot usefully be employed in classi- 
fication on account of the way in which two or more causes often blend, 
and on account of the difficulty of determining the cause in individual 
instances, even when there is only one. The causal classes that stand 
out clearly, or comparatively clearly, as euphemism, cover only a small 
part of the whole field. 

Comparing the causal point of view with the distinction between 
chpping and omission, we find that the two intercross, and that they 
could consequently be applied only in succession. It seems doubtful 
whether the causal classification is suitable even in a secondary place. 

The conditions of shortening are still less useful as a basis of classifi- 
cation, since the same short expression may be conditioned by different 
kinds of context on different occasions. The waiter's "sweet or dry, 
sir?" may be conditioned by perceptual context, the sight of the two 
bottles, or by mental context (see 10.17). 

There remains, then, in full agreement with the historical point of 
view previously adopted, only the formal side. Shortenings have to 
be classed according to their formal, morphological, character. As 
already indicated, the natural distinction is between the clipping of 
symbols, and the total omission of symbols. It is true that in this case, 
too, there are intermediate instances difficult to place, but that is a 
difficulty which can never be altogether avoided in semantic classifi- 

CUpping is the shortening of a unitary verbal symbol, as bus for 
omnibus, auto for automobile; omission is the total dropping of one or 


more verbal sjonbols forming part of a combination, as private for 
private soldier, fall for jail of the leaf. 

"Unitary" here means: forming a semantic unit. Such units are not 
necessarily written as one word, but sometimes as separate words, or 
hyphened: as Lady Superior, black-bird, in which the meaning of the 
combination is different from the combined meanings of the elements. 
Similarly, combinations may be written as separate words, or hyphened, 
or as one word: classroom, double-ended, private soldier. The written 
form is therefore no reliable guide to the semantic conditions, which are 
the basis of the distinction between unitary symbols and combinations. 
In a great many cases the distinction remains vague, but it is essential 
from the semantic point of view and has to be employed. (Cf. Fowler, 
MEU s. V. Hyphens, Nyrop IV 50 sqq., Wellander, Studien II 97). 

The importance of the distinction between clippings and omissions, 
and consequently of that between unitary symbols and combinations, 
is evident from the following circumstances. When omnibus is short- 
ened to bus, citizen to cit, brigantine to brig, spectacles to specs, peram- 
bulator to pram, and so on, no sense-change can be involved, because 
the stump is an entirely new word. When the stump happens to be 
identical in form with a previously existing word, as hack from hackney, 
cad from cadet, pop from popular concert, props from properties, and so 
on, the new word is rarely identified with the old word by linguistic 
feeling: they are apprehended as homonyms, and consequently there 
is no sense-change. Similarly when two shortenings give the same 
result: as rep from repetition and from reprobate. 

In omissions, on the other hand, the verbal symbol or symbols pro- 
nounced are, by definition, identical not only in form but also in fact 
with previously existing words, and sense-changes are bound to foUow 
in all cases where the shortening becomes habitual. 

The unitary symbols of the form black-bird present an intermediary 
type, which is discussed in 10.3. 

With regard to the further classification within the two main groups, 
the clippings that lead to a sense-change are so few in number, that no 
further classification appears necessary for my purposes. In the omis- 
sions, it seems expedient to continue the application of the formal cri- 
terion. I first distinguish omissions of one element in binary combina- 
tions from omissions in longer combinations. The former are much 
more numerous, and among them a further classification can easily be 

26o GUSTAF STERN" 10.21 

established. We shall find (10.41) that omission of the qualifier differs 
in several respects from omission of the headword, and I shall conse- 
quently make this the next division. Further groups are distinguished 
with regard to the syntactical nature of the words involved. 

10.22. Permanent and Contextual Sense-change. Delimitation of 
the Problem. The field of discussion has now been limited in several 
ways. The historical point of view involves a restriction to (i) clip- 
pings, and (2) omissions that may be described as individual. Again, in 
these two groups we have excluded all shortenings that do not lead to a 
sense-change. According to the plan of this book, a further delimitation 
must be made. I am only investigating instances of permanent change 
of the basic meaning of single words. This involves the exclusion of 
shortenings that lead to changes of relational meaning (some instances 
will, however, be mentioned in 10.3), and also all instances of contextual 

Clippings are of inferior interest to me, since they seldom lead to 
a change of meaning; see 10.21. 

In the case of omissions, the remaining portion of the expression is 
alwa^'^s formally and factually identical with a previously existing 
word, but the two words are not always identified by linguistic feeling. 
From a strictly genetic point of view the result of the shortening is a 
new word, as when private soldier is shortened to private, or Shrapnel 
shell to shrapnel. We have not, genetically, a sense-change of the words 
private and shrapnel. But to linguistic feeling, that is to say, for all 
practical purposes, we have a new meaning of the word private; with 
regard to shrapnel the matter is perhaps doubtful, at least at a later 
stage of the development. 

From such indubitable cases of permanent sense-change we have to 
distinguish shortenings which lead to contextual sense-change only, as 
Smith's for Smith's house, shop, rooms, etc., St. Paul's for St. 
Paul's Cathedral, Church, Chapel, Abbey, School; or five and six, which 
may refer to feet and inches, or to shillings and pence, according to the 
context (cf. Nyrop IV 59 — 60, quoting Darmesteter, La vie des mots 
44; Wellander, Studien II 102, 112) and so on. I suggest that these 
are contingent meanings, whereas private and shrapnel have acquired 
new "tied" meanings (cf. 4.25). The various referents for Smith's and 
St. Paul's do not form categories, as the referents of tied meanings do. 

It seems as if so-called typical shortenings give rise to contextual 


sense-changes only, while individual shortening leads to permanent 
changes. However, the material at my disposal is not large enough for 
deciding the question, and I have to leave it to future research. 

In the meantime, I shall have to restrict myself to individual omis- 
sions, as better known and more easily analysed. We have to discuss, 
then, (i) clippings, which seldom lead to sense-change and can therefore 
be briefly dismissed, and (2) individual omissions, which generally 
lead to sense-changes, and require a inore detailed commentary. 

10.3. Clipping. 

It has already been pointed out that clippings seldom give rise 
to sense-changes; as a matter of fact, I have not a single indisput- 
able instance. 

The shortening of citizen to cit, hrigantine to hrig, perambulator to 
pram, spectacles to specs, etc., which lead to the formation of totally 
new words, have to be left aside altogether. 

In other cases the stump word is identical in form with a previously 
existing word, as cad from cadet, pop from popular concert, props from 
properties, or two clippings may give the same result, as rep from repe- 
tition and reprobate. But it seems to be very rare indeed for the new 
formation to be identified with an older word, or for two identical 
stumps from different sources to be identified with each other. The 
words seem to be apprehended only as homonjons. A real association 
between them could scarcely be expected if there were not originally 
some similarity of meaning, which would naturally be a comparatively 
rare occurrence. Some special cases and types require discussion.^) 

The word Miss (on its origin, see 10.183) is first instanced in the 
NED as denoting 'a kept mistress, a concubine' (1645). A quotation 
from 1662 says . . . "Misse, as at this time they began to call lewd 
women". But almost simultaneously we find miss in its present use 
"prefixed as a title to the name of an unmarried woman or girl". It 
seems impossible for either of these two senses to have arisen out of 

*) For lists of clippings, and comments upon them, see L. Miiller 55 sqq., Witt- 
mann, passim, Greenough & K. 61 sqq., Jespersen, I,anguage 169 sqq., Sunden, 
EU. Words 50, 73, 184, Horn, Sprachkorper g, Kjellman 6 sqq., Camoy 240, 
Bergmann passim, Nyrop, Etudes 13 sqq., Weekley, Words 66, Fowler, MEU s. 
V. curtailed Words. The instances in the following sections are taken partly 
from these authors, partly from the NED, which has always been used to 
verify the statements. 


the other except through a comparatively long development. We have 
to assume that they represent two independent shortenings of mistress, 
in which word both meanings are much older, and due to French influ- 
ence. The clipping has thus given rise, if not to a sense-change, at 
least to two meanings of one word. I assume that it was apprehended 
as one word by seventeenth century linguistic feehng.^) 

An intermediate type is represented by German forms like der Korn 
for der Kornbranntwein. French la Saint- Jean for la fete de Saint-Jean, 
un premiere Paris for un billet de premiere classe a Paris, un poche for un 
violon de poche (Bergmann 18). In form they appear to be combinations , 
not unitary symbols, but in the shortening they are treated Uke unitary 
symbols, as is shown by the fact that they retain the gender of the 
dropped head-word. Compare instances like Viennese der Ring for die 
Ringstrasse, die Burg for das Burgtheater (Noreen 272), which are genuine 
omissions of headword; each element of the combination was still felt 
as an independent unit, and when the second element was dropped, the 
first element took its own gender. It is evident that der Korn 'corn 
brandy' is not identified by linguistic feeling with das Korn 'the corn'; 
there is thus no sense-change. (Cf. Wellander, Studien II, 4, 107, III 
viii, 3 sqq., 26, 28; Krueger, Arch. 107, 351 sqq., Bergmann 18, 19).^) 

In English, where gender is not indicated, we get no precisely parallel 

^) I may mention here the phenomenon termed by Sunden pseudo-ellipsis (Sun- 
d^n. Ell. Words 141 sqq.). Bob has been considered a hypochoristic shortening 
of Robert. But why should R- be changed into B-? Sunden points out that there 
existed in OE the proper names Boba, Bobba, Bobing, and in ME Bobbe, Bobin, 
Bobbet. Robert was introduced into England through the Norman conquest; a 
shortening of Robert gave Rob or Robbe, which are also instanced in ME. We have 
thus Bobbe and Robbe, of different origin, but both of them proper names. It is 
reasonable to assume that the two names were confused, and Bobbe apprehended 
as a short form of Robert. Sunden is able to show that a similar process is prob- 
able for William — Bill, Richard — Dick, Amelia — Emy, Edward or Edmund 
— Ted, Isabella — Tib, James — Jem, Jim, and others. We have here a double 
process: shortening plus phonetic associative interference (see 9.4). 

2) Wellander (Studien II 4 sqq.) attempts to make use of the indisputable 
shortening of Kornbranntwein to Korn 'corn brandy' to prove that Schirm 'um- 
brella' must be a shortening of Regenschirm. He has failed to notice that Regen- 
schirm is a species of Schirm, while Kornbranntwein is not a species of Korn, but 
of Branntwein. The latter should therefore be compared to der Ring, die Burg, 
as quoted above, and here the essential difference to linguistic feeling between 
clipping and omission at once appears in the change of gender. 


instances, but the number may show a cUpping: All Souls, as a shortening 
of All Souls' Day or All Souls' College, takes the verb in the singular, 
in spite of its plural form. 'Cf. German A Her Heiligen (Wellander, Stu- 
dien III 19) and see Jespersen, Grammar II 8. 93, 

The much-discussed case of French ne — pas and other similar ex- 
pressions will probably have to be considered as representing an inter- 
mediate type. French pas, point, rien, aucun, jamais, personne, from 
Lat. passum, punctum, rem, aliquem unum, jam magis, personam, are 
all originally positive words, and in all of them except rien the positive 
meaning persists by the side of the negative one. As typical for the 
group I take ne — pas. Jespersen (Negation 16) states that pas was 
first used with verbs of motion, in connection with which the primary 
meaning of the word was of course very much in its place. It functioned 
as an intensifier of the negation, and it had to submit to the same 
development as so many other intensifiers, the fading of the cognitive 
elements of meaning (cf. 14.531). The definite cognitive import 'step' 
disappeared, and there remained only the emotive element, the intensi- 
fying function. When this adequation was completed, ne — pas was 
capable of extension to other verbs which would not have been com- 
patible with it when pas still retained its original meaning. Ne — pas 
now meant 'not at all', and was, like not at all in English, a semantic 
unit, although written as two words. At this point of development, it 
becomes possible to drop the first member of the expression without 
detriment to its function as a negative. 

Simultaneously with, or subsequently to, the fading of the cognitive 
elements of ne — pas, a fading of the emotive elements set in, resulting in 
a weakening of the intensive force of the expression. This is also a case 
of adequation (14.532). 

Through these two processes we arrive at a state of things in which 
ne — pas is an ordinary negative; this is still the case in hterary French. 
Although, to the eye, two words, it is a semantic unit expressing a 
simple meaning.!) 

Like other unitary symbols, ne — pas can be subjected to cUpping, 
consisting in the dropping of the ne. The immediate causes of this 

^) Cf. Horn, Sprachkorper 7, and the instances quoted there; also Wundt II 
583, who states that "sich die Begriffsinhalte der verbundenen Worter so innig 
assoziieren, dass sie nur noch einen Begriff bilden." Wundt 's explanation is 
otherwise unsatisfactory. 

264 GUSTAF STERN 10. 3 

clipping may be phonetic. Jespersen (Negation 5) points out that 
when a negation is put at the head of a phrase or sentence, it is Uable to 
disappear owing to prosiopesis. A speaker begins to articulate, or 
believes he does, but in reality he does not produce any sounds until 
two or three syllables after the beginning of what he wanted to say, 
either because he does not at once use sufficient force of breath, or be- 
cause he does not place his vocal cords in the proper position. This is 
very common and may become a habit, especially in standing phrases 
(see 10.182). Moreover, before a vowel ne is reduced to n , and in 
careless speech, especially with a consonant preceding, the n is 
easily lost: ce nest pas becomes c'est pas. Functional and economic 
causes may contribute to the result (cf. Horn, Sprachkorper 93). The 
reason for dropping the ne, and not the pas, is probably to be partly 
sought in the slight phonetic body of ne, especially when its vowel is 
elided. Ne alone is retained in je ne sanrais le dire, and other traditional 
phrases; this is not a shortening of ne — pas but a continuation of the 
earher use of ne without re-inforcement. 

In this wa}', pas may finally come to mean 'not'. As Marty .states 
(Unt. 524 note i, and 669), the dropping of ne does not involve a 
sense-change. A change took place when pas, after connection with 
ne, acquired through adequation the meaning of a simple negation. 
The dropping involves a sense-change only if pas 'not' is identified, by 
linguistic feeling, with pas 'step', which is probably not the case. We 
may compare the dropping of ne with the shortening of automobile to 
auto (cf. lycrch. Negation). 

In English, an original ne — hutan has been shortened to hut: we nabhad 
her buton fif Jilafas and twegen fiscas > we have here hut jive loafs and 
two fishes (Horn, Sprachkorper 90), he nis but a child > he is but a 
child (NED hut 6). The immediate cause of the omission of the ne- 
gation is not quite certain. It is not impossible that influence from 
other uses of hut may have intervened. 

According to Kjellman (7) the tendency to clip words in French is 
practically entirely due to the lower strata of societ5\ In English, I 
think we may safely say that the tendency comes essentially from 
careless speech, vulgar or not.^) Wittmann states that student slang 

^) An instance at random: "It's frightf'ly awk for us, all this, said the young 
lady. I should never 've got there in time. You seen our show, 'Dat Lubly Lady'?" 
(Galsworthy, The Swan Song p. 25). The speaker is a chorus girl. 


consists in large part of clipped forms. Some of these seem to be made 
for convenience, others are humorous in intention (Wittmann 117).^) 

10.4. Omission. 
10.41. Preliminary Remarks. 

10.411. The Form of the Shortening. Contrasted Referents. As 
mentioned above (10.21) the omission in a binary combination may 
affect either the headword or the qualifier. It will be necessary to 
discuss the reasons for this variation. 

As a general principle, we may say that when no specific factors are 
involved, the natural way to shorten a binary combination of headword 
and qualifier, is to leave out the qualifier and use the generic word 
alone in a particularized meaning: red roses - — the roses (sc. those of 
which I am speaking, the red ones. On particular meanings, see 4.23). 

In some cases, this is the only possible way of shortening, as in the 
types fall (of the leaf) , to lay {to pledge) , to be short {of money) , in which an 
omission of the headword is out of the question, for obvious syntactical 
reasons (see 10.422 — 10.424). We can therefore leave these types 
aside at present, and confine the discussion to combinations consisting 
of an attributive noun or adjective with a nominal headword. 

If we are speaking about one thing, or referent, and using a name of 
the type red roses, the natural way to form a shorter name for the re- 
ferent is, as I have just stated, to employ the generic word alone, 
and it is a method of which we make a very frequent use. The explana- 
tion will be discussed below (10.412). If we are speaking of two or more 
referents, or if we are speaking of one referent, and thinking of one or 
more others, belonging to the same genus, their names often contain the 
same headword, as red roses and white roses, red and white roses. In 
such cases, it is obviously impossible to omit the distinctive qualifiers 
without running the risk of ambiguity. If the names are to be short- 
ened at all, it can be done only by omitting the headword. In English, 
the saving of .space and effort by this means is somewhat illusory, since 
the prop-word one must be used: the red ones, the white ones, but in other 
languages the method is more effective: les rouges, die Roten, de roda. 

1) The further classification of clipped forms is of no great interest to the pre- 
sent study. I refer, for points of view, "to Sunden (EU. Words 185 sqq.), who deals 
mainly with prosodic shortenings, Kjellman 20 sqq., Wittmann 118 sqq., and 
L. Miiller. 


Similarly, if we are speaking of referents named by combinations of 
identical qualifiers and different headwords, as German language and 
German literature, German language and literature, the only possible 
method of shortening is to leave out the common member, the qualifier. 

The form of the omission — the choice of method — is thus deter- 
mined by the nature of the contrasted or compared referent. If it 
belongs to the same genus — denoted by the headword — the head- 
word can be omitted; if it belongs to another genus, but has the same 
quality — denoted by the qualifier — then the qualifier can be omitted. 
But note that the conditions for the omission as such are always those 
described in 10.17: the presence of an adequate context, verbal, per- 
ceptual, or mental, which supplies the functions that should have been 
filled by the omitted word. 

The fact that another referent is contrasted with the one in question 
does not necessarily imply that the name of the contrasted referent is 
constructed in the same way (qualifier + headword); that is the case 
with red and white roses, and probably in most omissions; but not always. 
A private soldier (see also common soldier, NED s. v. common, adj. 12) 
is so named in contrast to officers; officers are also soldiers, although 
not expressly so named, and the omission of the headword was there- 
fore the only possible way of shortening. 

Nor is it necessary for the contrasted referent to be expressly men- 
tioned in connection with the one in question; that is to say, the contrast- 
ed referent need not appear in the verbal context, it may be present 
only in the perceptual or mental context, as in the following sentences 
(from a newspaper): For popularity none can approach cocker spaniels 
or Irish setters. Cockers, a long way in front of any others, are right on the 
crest of the wave at the moment. No other breed of spaniels is mentioned 
in the whole paragraph, but it would be misleading to shorten the name 
by leaving out the qualifier; there are other breeds, and the generic 
name would therefore be ambiguous. The writer's knowledge of 
other breeds (his mental context) is sufficient to determine the form 
of the shortening. 

Since the natural way of shortening, when there is no contrasted re- 
ferent present to the speaker's mind, is to leave out the qualifier, it 
follows that omission of the headword without contrasting referent is 
rare. I am speaking here of the actual shortening. If the shortened 
form becomes habitual, it is likely to be used as the name of the referent 


in any context, with or without contrast; that is the case with private 
{sc. soldier) 1). 

To sum up, the form of the omission — whether of headword or of 
qualifier — is determined, (i) if there is no contrast or comparison, by 
our speech habits, which mostly lead to omission of the qualifier; and 
(2) if there is a contrast or comparison, by the nature of the con- 
trasted referent, as described above. 

The omission as such is conditioned by adequate context in the 
usual way. 

10.412. Omission of Qualifier and Genus pro Specie. I mentioned 
above the use of a generic name for a species: roses for red roses. Are 
we to explain the use of roses = 'red roses' as a shortening of the fuller 
term red roses, or is it a case of genus pro specie with no direct genetic 
connection with the fuller term? In an occasional combination like 
red roses the latter explanation seems a priori more probable, but it 
might be otherwise in habitual combinations. 

If the generic term machine is used for a sewing-machine, both the 
explanations mentioned are theoretically possible. But if machine is 
used for a type-writer, it must be a case of genus pro specie. On the 
other hand, when fall 'autumn' is used for fall of the leaf it must be a 
case of shortening, since the meaning of fall here lies outside the original 
semantic range of the word, and a direct sense-change from fall 'act 
of falling' to fall 'autumn' is out of the question. It is thus only the 
type sewing-machine, where the generic name enters as an element into 
the name of the species, that requires further discussion. 

Compare the following paragraph from a newspaper: "Budapest, 
February 15. Although the opera ball has been cancelled this year 
owing to the economic crisis, the event of the carnival season is Itkely 
to be the ball given by the war prisoners, to be called 'One Night in 
Siberia'". In this short paragraph, we note the following conspicuous 
instances of words employed in particular meanings: the (Budapest) 
opera ball, the economic crisis (in Hungary), the (outstanding) event, the 
carnival season (this year in Budapest), the (Hungarian) war (i. e. Great 
War) prisoners (i. e. late prisoners who have returned alive and are now in 

^) Wellander ha6, in my opinion, misinterpreted the nature of the connection 
between syntactical parallelism and shortening (see the criticism in 10.5 below), 
"but he has the merit of having emphasized that there really is a connection of 
some kind, which must be taken into consideration. 


Budapest). A closer scrutiny would have to point out that cancelled , 
economic, season, likely, given, called, are all used in particular senses; 
that is to say, every word of any importance in the paragraph is used 
in a particular meaning, a meaning determined by context, mental or 
verbal, but chiefly the former. With the help of this context the reader 
is easily able to interpret the whole paragraph correctly. 

The passage is not peculiar in this respect: practically every word 
in every written or spoken sentence is thus used, not as denoting its 
whole referential range, but as denoting some particular referent or 
referents within the range (cf. 4.23). The use of words in particularized 
meanings is thus an extremely common phenomenon, and in most cases 
shortening cannot be adduced in explanation. On the other hand, the 
number of cases in which shortening can be proved (type: jail of the 
leaf, etc.) is ver}^ much smaller. If we go by the law of probability — 
when no other criterion offers — we ought to assume that our mental 
processes follow their most habitual lines (cf. Marty, Unt. 669). 

For my purpose — classifying sense-changes — we may argue as fol- 
lows. If we call a referent the International Labour Conference, or the 
equivalent in any other language, we have, eo ipso, classed it as a 
conference. If, in the sequel, speaking of the same referent, we denote 
it by the Conference, dropping the qualification, we cannot say that the 
word has changed its meaning, since the referent was from the begin- 
ning denoted by this generic name. Consequently, all instances of this 
type fall outside the scope of my stud}^^) 

In accordance with these principles, I include in the present chapter 
only indisputable instances of shortening, those in which the semantic 

1) If our purpose is to classify shortenings, the problem is not so easily solved. 
When the referent is, in the second or third instance of naming, called by the 
generic name, the conference , it might equally well have been called the meeting 
or the session. The last two must, in any case, be instances of genus pro specie,, 
and it seems reasonable to analyse the first in the same way, if no proof of 
shortening is available. The three nouns are parallel in various respects, and 
should, if possible, be explained in the same way. It seems better, in doubtful 
cases, to employ the explanation that is applicable to the great mass of words 
used in particular meanings, than to let a very large group follow the much smaller 
group in which shortening can really be proved. I may add that the explanation 
here recommended also covers the use of pronouns: we may refer to a previously 
mentioned conference as it (cf. Sunden, EH. Words 26, and Wellander, Studien 
II 76 sqq.). 


range of the generic name, after the shortening, falls outside its original 
range, and in which a direct sense-change is improbable. 

I think this principle may also apply to new referents. If Wellander 
is right in asserting that when umbrellas were first introduced into Ger- 
many, their name was Regenschirm, which was later on shortened to 
Schirm, then the term Regenschirm already implies that the new object 
has been classified as a Schirm of some kind (this is substitution, ch. 8). 
The subsequent use of Schirm alone is conditioned, as usual, by the 
presence of an adequate context, verbal, perceptual, or mental, which 
fills the functions of determinant that might also be filled by the word 
Regen-. It is probable that Schirm 'umbrella' should now be regarded 
as a unique use (see 14.85). 

One reason for the impossibility of upholding the distinction between 
omission of qualifier and genus pro specie, in cases where no sense-change 
definitely proves the former explanation to be correct, is the fact that 
our psychic processes are so variable, from person to person, and from 
occasion to occasion. A generic word with a particular meaning may, 
for one speaker, be a case of shortening, while for his hearer it is a case 
of genus pro specie. 

10.413. Further Delimitation. When we turn to omission of the 
headword, we are confronted by another set of problems; in many cases 
it is doubtful whether we should assume a shortening, a metaphor, or a 
permutation. Instances are the use of a painter's name for a painting 
{a Rembrandt), an author's name for a book [a Shakespeare), a place- 
name for an event that happened there [after Waterloo), etc. The mat- 
ter will be discussed again below (11. 7 and 13.2). I shall follow the 
same method as before, reckoning as shortenings only indisputable 
instances, of which there is a considerable number (see 10.43). Thus, 
when a four-oar boat is termed a four-oar, it is obviously a shortening. 

A qualifier and a headword mutually determine each other's meaning 
in a peculiar wa}-. When either is omitted, the determination must be 
provided by context in order that the remaining word ma}^ 
be correctly interpreted. In that respect there is no difference; 
the conditions of shortening are essentially the same in both cases. 
That is true also of the causes of shortening. It is the syntactical func- 
tion of the omitted word that is different, as well as the syntactical 
results of the omission. 

When the qualifier is omitted, the headword retains its syntactical 


function without change. But when the headword is omitted, the 
qualifier, remaining alone, receives a new syntactical function. An 
attributive noun presents no difficulty; it simply takes over the functions 
of the headword; a how window becomes a how. Attributive adjectives 
are substantivized: private soldier > private. This, too, often occurs with- 
out any difficulty. Omission of noun plus preposition is rare in 
English, but follows the same lines (see 10.433). 

The omission of a verb qualified by an adverb naturally presents 
syntactical difficulties, since an adverb could not easily take on verbal 
functions. Apparently it is only in English that adverbs of direction 
are actually used and inflected as verbs: to down, to up; I outs into the 
street again (Kipling, Barrack-room Ballads, Tommy) .i) In German, 
for instance, such a development is scarcely imaginable (Wellander, 
Studien III 14). In the English instances, omission of a verb is gener- 
ally assumed. Various verbs might stand in such a position, and we 
should therefore have to assume typical shortening (Cf. Krueger, Arch. 
108, 121, ly. Miiller 48, Sunden, Ell. Words 20). Wellander is doubtful 
about the matter, and quotes Paul for another explanation (Studien 
II 124, Paul, Prinz. 320). In English, at least, I should prefer not to 
assume shortening. It was possible, from the beginning, to say he is 
down, he is up, etc. A further development would be, / must up, he will- 
down, in which we still, originally, have the adverbs, but in a position 
generally held by an infinitive. Since an English infinitive has no 
distinctive ending, the adverb could easily come to be apprehended as 
an infinitive, and other verbal forms be analogically deduced. The 
circumstance — pointed out by Miiller — that adverbs in -ly are not 
thus used, may perhaps be interpreted in favour of the explanation 
given. That ending is typical of adverbs, but very rare in verbs, and it 
is not unreasonable to assume that it might prevent the adverb in com- 
binations with he is, he will, he must etc., from being apprehended as a 
verb. It is less easy to say why the ending should have prevented the 
omission of a copula, as assumed by the theory of omission. 

There is, however, a third explanation, which would bring these cases 

^) Often, indeed quite recently, I have sat 
Sceptred and orbed the absolutist throne, 
Have upped this favourite, downed that other one. 
This absolutely good, that utterly bad. 
Robert Graves, Modern Poetry, Adelphi, vol. II, p. 290, 


into a large and well-known group, viz. to explain them as instances of 
primitive shortness {urspriingliche Kiirze, see 10.5). Sentences without 
a verb are common from the oldest periods of our languages. Brugmann 
quotes a number of different types in K. vgl. Gr. § 941. In questions, 
orders, and exclamations, it seems to be especially common to leave 
out the copula; the practice could no doubt easily be extended to other 
types. Wellander (Studien II 124) quotes numerous German and Swed- 
ish instances of the type Ich in voller Carrier e nach Haus! Jag upp ur 
sangen och ut i farstun. The problem is of some interest, and would 
require further investigation. 

10.42. Omission of Qualifier in Binary Combinations. 

10.421. (Attribute) -\- Noun. The attribute may be a noun or an 
adjective. I have no satisfactory instance of the omission of a nominal 
attribute. [Neat) cattle, and (police) force are adduced, but I am very 
doubtful about them. When these referents are so named, they are, eo 
ipso, classed as cattle or force, respectively, and the use of the generic 
word alone may just as well be a case of genus pro specie. Similar 
observations apply in many cases to Wellander's instances of Bestim- 
mungsellipse (Studien III 127 sqq.) He quotes, among other words, 
Zeichengeber and Zeichenempf anger, shortened to Geber and Empfdnger. 
The terms are used by Martinak in his Psychologische Untersuchungen, 
and they are probably coined by him for his own purposes, even if they 
had been used previously by others. I am uncertain whether Geber 
could be genus proximum to Zeichengeber ; if not, it is a genuine case 
of shortening; but I suppose that Empfdnger is gen.prox. to Zeichenemp- 
f anger. 

Authentic cases of the omission of an adjective attribute are also rare 
in Hnglish. Judging by the quotations in the ISTED, corn, in its Amer- 
ican sense of 'maize', is a shortening of Indian corn (Franz 31). Conceit 
'good opinion of oneself may be a shortening of self-conceit or of conceit 
of oneself (NBD s. v. conceit sb. 6). Libel 'a leaflet, bill, or pamphlet 
posted up or publicly circulated; esp. one assailing or defaming the 
character of some person', is short for famous libel, from Law Latin 
libellus famosus (NBD. Cf. Weekley, Words 42. French instances in 
Bergmann 18). 

The omission of a genitive seems to be instanced in the following 
cases. Knight originally signified 'a boy or lad employed as an atten- 


dant or servant; hence, by extension, a male servant or attendant of 
any age'. With a genitive or a possessive pronoun it canie to mean 'a 
military servant or follower (of a king or some other specified superior)'. 
According to this view, the king's knight, originally 'the king's servant', 
would take on the meaning of 'a military servant of the king,' and 
since these were often men of noble birth, it came to be applied to men 
of higher rank. Through the dropping of the genitive, knight alone 
received the same meaning. Similarly, wife 'woman' with a genitive 
or a possessive pronoun comes to mean 'somebody's woman', that is 

10.432. Noun + [Prepositional Phrase). Fall 'autiunn', from fall 
of the leaf has already been quoted (cf. Sunden, Ell. Words 42). Imp, 
originally 'young shoot of a plant or tree, sapling' etc. acquired the 
meaning 'scion (esp. of a noble house), offspring, child'. The word was 
then used in phrases like imp of serpents [Hell, death, damnation, the 
devil, etc.), and, with omission of the qualifier > 'a little devil or demon, 
an evil spirit' (NED). Flour is the same word as flower. Originally it 
signified the 'flower' or finest quality of meal; in all the early quotations 
is added: of wheat, of barley, of rice, of meal. From 1691 the NED 
quotes: "Milk, Water, and Flower, seasoned with Salt. . . are rare Foods 
for Them" . Here there is apparently no qualification in the verbal 

Two pair of stairs was shortened to two pair; the earliest instance in 
the NED is from Dickens (1844): "a spacious room on the two-pair 
front". Mail [of letters) originally denoted a 'bag or packet of letters 
or despatches for conveyance by post, hence the letters so conveyed'. 
The Mails for Calais (1684, NED). The shortening is followed by a 
permutation, from 'letter-bag' to 'letters'. A railway train was origin- 
ally a train of carriages, or waggons. (NED. Cf. Carnoy 114 and 
128, Bergmann 21 sqq.). 

10.423. Verb -f [Obfect or Complement). The instances of this 
type are comparatively numerous. 

To put to is shortened from to put the horses to. The NED shows the 
following series: To put the horses to the carte (1565); / bid him get the 
horses put to (1768); . . . that James should put to for such a purpose 

*) The earliest instance in the NED runs: "Kalues fleis, and f lures bred, and 
buttere" Gen. a. Ex. 1013 (c. 1250). Here the word bread is no doubt to be regarded 
as supplying the necessary qualification. 


(1815). It may be suggested that the passive construction in the 
second quotation serves as intermediary in the shortening. To lay 
'to stake, wager' is shortened from to lay to -pledge: jerne he wile ^e 
hidde and preie, ^at ^u legge J>e cupe to pleie ( = pledge; 1300); later 
on we find the shortened phrase: Of Charlemeyn ne his ferede nabbep 
pay non help, y legge (1380). Other instances, which may be similarly 
traced, are to dissolve {Parliament), to leave {school), to open {fire), to 
pop {the question), to shake {hands), to strike {colours), to take {effect), 
to hang out {one's sign-hoard), to knock off {work), to break {an officer), 
to call {to the bar), to hold forth {the word of life; cf. Franz 33), to stand 
{candidate) for a club, to score {success), to recruit {one's strength), to 
give notice {to quit, to leave), to retire {to bed, to rest). (Cf. Sunden, Ell. 
Words 40 sqq., Krueger, Arch. 107, 361 sqq., Bergmann 38, Wellander, 
Studien III 52 sqq., 133 sqq., Stocklein 68). 

10.424. Adjective or Participle -\- {Qualifier). Convincing instances 
are rare. To be short {of money) is adduced by Krueger (Arch. 108, 109). 
There is, however, an alternative explanation. Short is instanced in 
the NED from late ME in the sense 'not coming up to some standard 
of measure or amount, inadequate in quantity', as short measure, short 
weight, etc.; short payment is given from 1681; the money was short 
by jd., from 1753. The transfer of the adjective from the article that 
is wanting, to the person that is wanting in the article, would be a 
transfer of epithet of a very common type. There are a number of 
instances illustrating it in Stern, Swift (271 sqq.). Paul makes an ob- 
servation (Prinz. 320) on "die schon vor der Entstehung aller formellen 
Elemente der Sprache vorhandene und immerdar bleibende Fahigkeit, 
die Beziehung in welche zwei Begriffe im Bewusstsein zu einander 
getreten sind, mag dieselbe nun eine unmittelbar gegebene oder eine 
durch andere Begriffe vermittelte sein, durch Nebeneinanderstellung 
der Bezeichnungen fiir diese Begriffe auszudriicken". The transfer of 
an adjective from a thing to a person who stands in some relation to it 
must have been common in speech from early times. Similar observa- 
tions would apply to Wellander 's instances rot, bleich {im Gesicht), 
teuer, billig {im Preise). (Studien III 240. Cf. 13.3). 

Engaged 'fiance' may be a shortening of a longer phrase. The 
instances in the NED, s. v. engage 4 b, are not entirelj^ decisive. The 
first of them seems to imply a preceding context that determines the 
meaning of engaged. The other two instances have both the phrase 

Goteb. Hogsk. Arsskr. XXXVIII: i. 


engaged to a young lady {a young man) . Possibly this specialized meaning 
of engaged is as yet contextual only. 

10.43. Omission of Headword in Binary Combinations. 

10.431. Attributive Noting (Noun). This is a very common type. 
If the two links in the combination are intimately fused, we have in 
reality not an omission but a clipping (cf. 10.3: Kornbranntwein and 
All Souls), but in English there is often nothing to show how we should 
explain an individual instance. 

The following are typical: Lancet for lancet arch, or light, or window; 
bow for bow-window, brake for brake-van, bugle for bugle-horn, bull's-eye 
for bull's-eye lantern, porter for porter (or porter's) beer, staple for staple 
ware, the Underground for the Underground Railway, buttonhole for 
buttonhole flower, saloon for saloon car or carriage, safety for safety 
bicycle or lock, stock for stock-gilli flower. 

The qualifier is often a proper name: gladstone {bag), bramah {lock), 
bristol (stone), the Cape (Colony), the Carlton (Club), shrapnel (shell). 
Port for portwine should perhaps be reckoned as a clipping. (See 11. 331, 
and Franz 29 sqq., Sunden, Ell. Words 23, Krueger, Arch. 107, 361).^) 

According to the NED, the use of the article with names of ships, 
hotels, etc., is due to an appellative noun being omitted: the (ship) 
Nicholas, the Mermaid (Tavern). 

10.432. Adjective + (Noun). In this type, as in the preceding, 
we must sometimes suspect that the shortening is not an omission, but 
a clipping. The instances are very numerous. 

Private is a shortening of private soldier; the simplex is at first unin- 
flected, even in the plural, but it is inflected in a quotation from 1810 

^) In French, the attributive noun mostly conies after the headword, and 
omission of the latter seems pretty frequent: un chat Angora is shortened to 
un Angora (Bergmann 22). 

Of Wellander's instances (Studien III 11 sqq.) some are shown by their gender 
to be clippings (as eine^ Bock tmn£ek, das Mittag for das Mittagessen, Swed. vdxtet 
for vdxtmargarinet) . Others seem to be instances of naming: Pour le mdrite is 
the inscription on an order, and the order is naturally so named. Cf. Swed. 
Liiteris et Artibus, denoting a medal with that inscription. Similarly Schloss 
Johannisberg and Kupferberg Gold are names printed on the labels of these wines, 
and we consequently name the wines thus, without further additions. Books 
have been thus named from their authors, as OE. sea boc, he man Orosius nemned. 
Cf. 13.24 on permutations of this kind. — Other instances of Wellander's are no 
doubt genuine omissions. 


(NED). Other instances are general {housemaid; or, but probably only 
contextually, omnibus, that is to say, an omnibus belonging to the 
lyondon General Omnibus Co.), a blue {-stocking; clipping?), the blues for 
the blue devils, a constitutional {walk), a commercial {traveller), bay {-ant- 
ler), the main {sea), a natural {fool). Scout (F. escoute) is originally 'the 
action of spying out or watching', later 'one sent out ahead of the main 
force in order to reconnoitre' (1555); this may be a shortening of scout 
watch 'sentinel, spy', which is instanced from 1350 (NED). Periodical, 
daily, weekly, monthly (sc. paper or review), are now used as nouns; 
epic may be from epic poem. The earlier epidemy has been supplanted 
by epidemic, from epidemic disease. It seems probable that penal 
{servitude) , hard {labour), casual {labourer, pauper, or patient), perhaps 
also private {soldier), are due to graphic shortening. Otherwise all the 
different causes of shortening may be found in this type.^) 

We sometimes find occasional omissions, as it's no earthly (sc. use; 
Galsworthy, Silver Spoon p. 22). 

According to the NED, the use of the definite article before names 
of languages implies that the name is "consciously elliptical": a new 
translation from the Hebrew. Sweet (NEG 2040) states that in a book 
translated from the German "the article seems to suggest 'the German 

10.433. {Noun + Prep.) + Noun. The type is probably rare in 
English. Franz (28) states that exchange = Bill of Exchange, is a 
shortening of the latter combination, but the evidence available in the 
NED does not seem conclusive. However, the two expressions may 
undoubtedly be synonymous. Franz also thinks that a sherry is a 
shortening of a glass of sherry, but this is probably wrong. Many names 
of substances and liquids are used for varieties or qualities of the sub- 
stance: a tea, a sherry = a kind or quality of tea, of sherry. This is an 
excellent sherry, or try this sherry might be said by a person offering a 
glass of sherry; sherry would then be used, in reality, of a glass of the 
wine, and permutation would account for the change to that meaning.^) 

^) The type is very common in German and French; see Wellander III 10, Franz 
30 sq., Bergmann 10 sqq., 29 sqq., Nyrop IV 59 sqq. Krueger, Arch. 107, 361, 
369, Stocklein 65 sqq. Carney 244. 

*) I may note here the freakish (intentional) shortenings hot with 'hot spirits 
and water with sugar', and cold without 'brandy or spirits in cold water without 


In French this type is common: tm bateau a vapeur is shortened to 
tin vapeur, and so on. (See Bergmann 9 sqq., 21 sqq.). 

10.44. Omissions in Longer Combinations. Sunden states (Ell. 
Words 57 sqq.) that the Germanic languages do not favour three- 
linked compounds, and still less four-linked: twelfth-tide or twelfth-night 
cake is shortened to twelfth-cake. The shortening is probably due to 
analogy, as Sunden says. In many cases where the two-linked com- 
pound seems illogical, I think we have nevertheless to be cautious about 
assuming shortening. As pointed out by Paul in the paragraph quoted 
10.424, the relations between the elements of a compound may vary 
within the widest limits, without any connecting link being necessary. 
Sunden adduces news-boy "for the logically correcter newspaper-boy" . 
But the latter word is not given by the NED. Moreover, the earliest 
instance of newsboy is from 1801, and at that time news was not seldom 
used = 'newspaper'; there were also the compounds newsbringer, news- 
bearer, newscarrier , etc., which might serve as patterns for the forma- 
tion of newsboy. It seems almost certain that we should not assume 
shortening. Nor is there any evidence in favour of bull-pup being an 
abbreviation of bulldog-pup. Probably bulldog was felt as a compound; 
since dog and pup are correlatives, the latter was easily substituted for 
the former. 

Horn (Sprachkorper 5) quotes other instances of a similar type and 
explains them as shortenings. But even if fire-office and fire-insurance- 
office exist side by side, that is no proof that the shorter compound has 
developed out of the longer one. The compounds with fire- in the NED 
present such a variation of relations between the two elements, that a 
formation like fire-office can have nothing strange about it, and to have 
recourse to shortening in order to explain it is not theoretically necessary. 
How the term really arose is another matter, and the material in the 
NED is not sufficient for deciding. With regard to Whitsun, the NED 
gives a more plausible explanation than Horn. Whitsunday, originally 
Whitsunday, was analysed as whitsun-day (cf. Christmas day, Easter 
day), which naturally gave rise to whitsun eve, whitsun ale, etc. On the 
other hand, Whit Sunday, Whit Monday, etc., gave rise to Whit Week. 
Thus there is nowhere any shortening. Why should Sonnabend neces- 
sarily be < Sonntagabend as Horn, 1. c, Miiller, and Giintert, Grundfragen 
32, affirm? We have Oster Tag and Oster Abend, Weihnachtstag and 
Weihnachtsabend; why not Sonntag and Sonnabend? 


With regard to certain Swedish instances reckoned by Noreen as 
elliptical, Sunden (EHipsbegr. 227) prefers to explain them as analogical: 
te{kopps)fat, arm{ljus) stake, tretton{dags)afton, kam{ulls)garn. Sorne of 
these have never existed as three-linked. It is no doubt right that three- 
linked cornpounds are as far as possible avoided, but that does not 
prove that the two-linked synonyms are necessarily shortenings. We 
must know the history of each word in order to pronounce upon it. 

An extreme case of these omissions is furnished by expressions like 
Pater Noster, being the name of the prayer thus beginning. Dirge is 
from lyat. Dirige, which similarly is the first word of a prayer; Ave 
and Creed (< Credo), Tedeum, Requiem, Reseda (from Reseda morbis, 
the beginning of a formula used as a charm when applying the plant to 
the reduction of tumours (NED), cf. Hirt 90), are other instances. 
(Cf. Weekley, Words 4 sq., 69, Sunden, Ell. Words 44, Wittmann 116). 

10.5. A Note on E. Wellander's Theory of Ellipsis. 

10.51. Wellander (Ellipsteori, and Studien II — III) has proposed a 
theory of ellipsis fundamentally different from the one I have adopted. Some 
remarks on it are necessary in order to show why I consider it untenable. 

Wellander first enumerates, under the heading of nicht-elliptische Wortauslassung, 
the following groups; Euphemismns (cf. 11,8), Aposiopese, Emphase (cf. 14.83), 
prosodische Kiirzung (cf. 10.182), graphische Kiirzung, graphisch bedingte Wort- 
iinterdruckung (cf. 10.183), urspriingliche Kiirze (cf. 10.13. i). ^^^ semasiologisch 
bedingte Wortreduktion. With the exception of the last, I shall have to pass over 
all these groups, although there are many statements that challenge criticism. 
The point of greatest interest are the two classes of partial and total ellipsis^ 
together with the Wortreduktion (semantic reduction), which is closely allied to- 
them. I shall follow Wellander in confining the discussion mainly to binary com- 
binations consisting of adjective and noun. 

10.52. Definition of Ellipsis. The term ellipsis is restricted by Wellander to- 
omission in parallel members of a sentence, of the type helles und dunkles Bier, 
deutsche Sprache und Literatur. His definition (Studien II 89) runs: "Die Ellipse 
konnte also jetzt genauer definiert werden als die syntaktische Erscheinung, dass 
in korrespondierenden zweigliedrigen Gefiigen mit einem Gliede gemeinsam 
dies gemeinsame Glied erspart wird, zunachst in einem Gefiige, partielle Ellipse, 
dann eventuell in beiden, totale Ellipse .... Die partielle Ellipse ist die nur 
einmalige Setzung eines doppelt oder mehrfach bezogenen Redeteiles, die totale 
Ellipse ist die bei haufigem Gebrauch eintretende (sit venia verbo!) keinmalige 
Setzung desselben Gliedes. Kiirzer konnte die Ellipse gefasst werden als die 
teilweise oder ganzliche Auslassung des Gemeinsamen in korrespondierenden Ge- 

10.53. Partial Ellipsis. What Wellander calls partial ellipsis, I have defined 
as an alternative method of formulating a certain mental content, more frequent 

^yS GUSTAF STERN 10. 53 

than the corresponding longer scheme: helles oder dunkles Bier is more frequent 
than helles Bier oder dunkles Bier (cf. 10.13, red and white roses). Wellander, 
too, regards this as a clear case of "einmalige Setzung mit zweifacher Beziehimg", 
and asserts definitely that "von einem "Auslassen" des Wortes Bier bei helles 
kann, historisch betrachtet, allerdings gar keine Rede sein" (Studien II 87). The 
logical conclusion would be that "partial eUipsis" is a type oiur spriingliche Kiirze, 
but in order to bring it under the heading ellipsis Wellander postulates a "Ver- 
schiebung der syntaktischen Gliederung" through which helles alone assumes the 
meaning helles Bier; after this process, the impression of an omission is said to be 
in many cases justified (1. c. 88). This strikes me as a mere logical spectdation, 
and it is not even applicable to all of Wellander's own instances. In the phrase 
er vertrauele mir erstlich seine Sau, zweytens seine Ziegen und zuletzt seine gantze 
Heerde Schafe (Studien III 48), the Verschiebung would imply that seine Ziegen 
alone acquires the meaning 'er vertraute mir seine Ziegen', which is absurd. 
Wellander has, I think, been misled by the circumstance that Helles, standing 
alone, may mean 'helles Bier', but he has not noticed that this is due to other 
processes: the adjective is substantivized, and then undergoes adequation, acquir- 
ing a unique meaning (cf. 14. 85). The latter, at least, is not the effect of short- 
ening, but is a separate, independent development, which is not found in instances 
like deutsche und englische Literatur. Adequation seems to have occurred also in 
several of the instances adduced by Wellander in Studien III. They are therefore 
unsuitable for illustrating the effects of shortening alone. 

When the theory of Verschiebung breaks down, we are left with the conclusion that 
partial ellipsis is a type of ursprUngliche Kiirze {no omission), and total ellipsis a 
type of omission. It seems scarcely appropriate to bring these two under one 

10.54. Total Ellipsis. Total ellipsis is said to arise out of partial ellipsis "bei 
haufigem Gebrauch", and only in cases of parallelism (cf. 10.52). It may effect 
the headword or the qualifier of a binary combination. Typical instances are 
Helles oder Dunkles <C helles oder dunkles Bier; and Schirm und Mantel <C Regen- 
schirm und -Mantel. 

I think the matter wiQ become clearer if we put the groups in tabular form. 

1. With parallelism. 

a. Omission of headword = Wellander: Total Ellipsis of headword 

b. Omission of qualifier = ,, Total Ellipsis of qualifier 

2. Without parallelism. 

a. Omission of headword = ,, Analogy 

b. Omission of qualifier = ,, Semantic reduction 

It is obvious that these four classes exhaust the possibilities of omissions in 
binary combinations. The question at once arises how they are related to the 
nicht-elliptische Wortauslassungen, which are not restricted to other than binary 
combinations. Wellander would reply that ellipsis is due to syntactical causes, 
while the other classes have euphemistic, graphic, prosodic, or semantic causes. 
He mentions several times the difficulty of applying these distinctions, especially 
that between ellipsis of qualifier and semantic reduction. The latter is nowhere 


precisely defined, but it seems to embrace omission of qualifier in non-parallel 
combinations, as when die internationale Arbeitskonferenz is first thus called, 
and later on referred to merely as die Konferenz (Studien II 76. I prefer to 
regard this as a case of genus pro specie; (cf. 10.412). 

To illustrate the difficulties in question, Wellander quotes the use of Maschine 
for Ndhmaschine. This may be a case of semantic reduction; secondly, if Ndhma- 
schine has been contrasted with Ndhnadel, etc., it may be ellipsis; thirdly, it may 
not be a shortening at all, but a "Gebrauch kat' exoken", or, as I prefer to say, 
specialization or the unique use (Studien II 84). With regard to the first and 
third of these explanations, Wellander concludes that "theoretisch sind die semasi- 
ologisch bedingte Reduktion und der Gebrauch kat' exoken klar geschieden, in 
der Praxis dagegen ist es schwer, ja oft unmoglich, sie scharf auseinanderzuhal- 
ten". With regard to the distinction between reduction and ellipsis, Wellander 
admits that "es bleibt schliesslich nichts anderes iibrig, als beide Erscheinungen 
in einem Zusammenhang zu behandeln" (Studien II 86), and "es hat deshalb 
keinen Zweck, bei der Besprechung der Einzelfalle eine Scheidung durchfiihren 
zu woUen, die nur ein theoretisches Interesse beanspruchen konnte" ('Studien 
III 51). In other words, we find that, when Wellander comes to handle his mate- 
rial he makes the heading Ellipse cover, (i) partial ellipsis, involving parallelism 
but no historical shortening, (2) total ellipsis of qualifier, involving parallelism 
and historical shortening, and (3) semantic reduction, involving shortening^) but 
no parallelism. In practice, then, as distinct from the theoretical argumentation, 
Wellander does not make the parallelism decisive tor his class of Ellipse, and the 
reader is somewhat surprised that so much trouble should be taken to establish 
classes which cannot be used in classifying, and the difference of which "nur ein 
theoretisches Interesse beanspruchen konnte". 

Omission of headword without parallelism is rare (I have explained the reason 
above, 10.41 1), but I do not see why it should be attributed to analogy. 

10.55. Total Ellipsis <C Partial Ellipsis? Total ellipsis is said to arise out 
of partial ellipsis "bei haufigem Gebrauch". The series helles oder dunkles Bier 
^ Helles Oder Dunkles may look tempting, but a varying degree of completeness 
is no infallible criterion of historical development (cf. 10.13 and 10.14). I have 
already noted that Helles and Dunkles illustrate a shortening plus a subsequent 
adequation. In German and English Literature, or German Language and Literature 
there is no such process. We might very well find in a catalogue of books the 
following headings: IV. Germany, i . . . 2 . . . 3 . . . 4. Language and Literature. 
I do not see why this should not be immediately understood as referring to the 
relevant German literature, even if the combination were perfectly new to the 
reader. We would then have total ellipsis without previous partial ellipsis. Could 
not a child, holding out two apples, ask a friend, "Willst du den Roten oder 
den Weissen haben?" without any previous "haufiges Gebrauch" of the 
phrase rote und weisse Apfel? It seems to me that the only thing required 
is an adequate context (cf. 10.17). It is indeed probable that in some 
phrases the longer scheme has actually preceded the shorter one, in other words. 

According to Wellander. I regard these cases as only sometimes shortened. 


some phrases show individual shortening, but we do not know to what extent 
that is the case. In view of the immense variability of the psychic processes it 
is unlikely to be a general rule. The process may vary for different speakers 
with regard to the same phrase, and any attempt to prove a rule seems 

10.56. The Parallelism. It is obvious that an external criterion of ellipsis 
like "korrespondierende Gefiige" would be very convenient and easy to, handle. 
The question is whether Wellander has interpreted its significance correctly. 

I note first that, as explained above (10.41 1), the verbal (sjmtactical) parallel- 
ism is only the reflection of a contrasting or comparison of referents, and that such 
contrasting may have precisely the same morphological and semantic effect 
even if it is not expressed in words, but only exists in the form of mental context. 
Wellander has failed to notice the secondary nature of his criterion. 

Further, Wellander neglects the indispensable discrimination between the 
causes, the conditions, and the form of shortening. Especially the two former 
are confused, the terms Griinde, Bedingungen, and Ursachen being used promis- 
cuously. Wellander regards parallelism as a cause of omission: "Das Ausfallen. 
eines Gliedes der Wortgruppe beruht nicht auf der Art seiner Bedeutung, sondem 
auf seiner syntaktischen Stellung als gemeinsames Glied" (Studien II 95). The 
expressions syntaktische Griinde and syntaktisch bedingt often recur (for instance 
II 104, III 51). It is, however, fairly evident that the causa movens of a linguistic 
process must be a dynamic factor, whereas the syntactical configuration of a 
phrase is a formal quality (cf. 10.18). 

Wellander quotes among other instances; "Wer sonst vielleicht') die Zusam- 
mensetzungen Regenmantel und Regenschirni gebrauchen wiirde, der spricht, 
wenn von beiden die Rede isfi) getrost von Schirm und Mantel: Und ausgerechnet 
bei diesem Wetter muss ich den Mantel und Schirm. vergessen" (Studien III 48). 
I am unable to comprehend why a speaker should not say, quite as getrost: Und 
ausgerechnet bei diesem Wetter muss ich den Schirm (or den Mantel) vergessen t 
The total omission of Regen- is of course conditioned by perceptual context, 
clearly referred to in the words bei diesem Wetter; with or without the parallelism 
this context is equally adequate, and equally necessary (cf. 10.17). Similarly 
with ellipsis of headword; Wellander argues (Studien III i — 2) that if we are 
speaking of Jakob and Wilhelm Grimm, we can easily leave out the surname and 
call them merely Jakob and Wilhelm (cf. 10.17). It seems to me that if we are 
speaking of Oliver Cromwell, not mentioning any other Cromwell or any other 
Oliver, it is quite as easy to use either name: we may call him Oliver or we may call 
him Cromwell. The importance of the parallelism is greatly exaggerated by Wel- 
lander. As I have shown (10.41 1) it is only the form of the shortening that is 
determined by the parallelism (more correctly: the contrasting). 

Even if the parallelism is neither a cause nor a condition of shortening, it might 

1) If this vielleicht is taken seriously, it simply breaks down the whole 
distinction that is being established. But in the sequel Wellander generally 
argues as if there were no vielleicht. 

2) My italics. 


still be useful as a criterion, if a definite type of shortening was found to be per- 
manently correlated to it. The instance of N dhmaschine shows that this is not 
the case, and in practice, as Wellander admits (10.54), the results of shortenings 
with and without parallelism (ellipsis and semantic reduction) cannot be dis- 
tinguished. Since it is the results, i. e. the shortened expressions themselves, that 
are the point at issue, it seems inevitably to follow that we have no use at all for 
the parallelism in classifying shortenings. 

10.57. Conclusions. The result of this analysis is (i) that the class ellipsis 
comprises two disparate types; one of them, partial ellipsis, is not an omission at 
all and should not be bracketed with total ellipsis; and (2) that the shortenings 
effected through semantic reduction and ellipsis of the qualifier are often indis- 
tinguishable. On both accounts, the class that Wellander has sought to establish 
through the criterion of parallelism breaks down. We cannot avoid the conclusion 
that parallelism is not essential for shortening. 

Wellander has, I think, confused classification and analysis. For the latter, 
every circumstance that has a bearing on the problem in hand should be noticed 
and explained. But if two or more of the processes involved give absolutely iden- 
tical results, they are useless for discriminating types among these results — 
the shortenings. For a classification of linguistic phenomena, valid for a whole 
speaking community, we cannot utilize processes that vary from speaker to 
speaker, but only those that are constant concomitants of one or the other 
group of phenomena, and of that group alone. 

Although admitting the occurrence of specialization of meaning not only as the 
result of shortening, but also through "Gebrauch kat' exoken", Wellander does 
not seem to have realized the enormous frequency of specialization without short- 
ening. Having confined his investigation to one class, he is unable to weigh the 
claims of alternative explanations, and so is induced to include among Auslass- 
ungen and Ellipsen many instances that should have been otherwise interpreted. 
The groups that result are so heterogeneous that tenable definitions cannot be 
formulated, and the failure of the classification when applied to the linguistic 
material follows inevitably. 

The inadequacy of a theory does of course not prevent explanations of details 
from being acceptable, and my numerous references to Wellander 's writings show 
us to be in agreement on many points. 



11.1. Theoretical Discussion. 

11.11. General Characteristics. The intentional selection of terms 
by the speaker rnay involve a sense-change. Sonietimes the referent 
is a new invention or discovery, calling for a nanie. The speaker may 
either invent a new name or transfer an old name to the new referent. 
Sometimes the referent is not new, but for some reason its habitual 
name does not appear to the speaker as the best way of denoting it in 
the actual situation. In this case, too, the speaker may invent a new 
name or transfer an old name to the referent. 

The intentional naming of a referent, new or old, with a name that 
has not previously been used for it, will be called nomination}) 

The invention of a new name does not necessarily involve a sense- 
change, since the word in question did not previously exist, but through 
secondary processes a sense-change may follow. The transfer of an 
old name to a new referent involves a sense-change, at first in the indi- 
vidual instance, and later on, if the practice becomes habitual, as a 
permanent addition to the semantic range of the word. It is only 
permanent sense-changes that are of interest for the present study. 

11.12. Types of Nomination. As indicated in ii.ii, I make a 
distinction between words that are new coinings, and words that are 
intentionally transferred from one referent to another. The former 
type I shall call intentional naming. It is of no great importance for 
my purposes, but I shall have to say something about it on account 
of a secondary sense-change that may sometimes follow. 

^) Cf. the German term Namengebung. Collinson, Mod. Lang. Review 20, 1925, 
p. 103, suggests the term onomathesia. 


Among the remaining types, in which a name is intentionally trans- 
ferred from one referent to another, I make a first distinction with 
regard to the psychic causes that lead to the shift. They may be pure- 
ly cognitive, or they may be more or less emotive. The former give 
rise to intentional transfer (non-figurative), the latter to figures of speech. 

The cognitive causes of nomination are to be sought in the striving 
for a better adaptation to the symbolic and communicative functions 
of speech, the speaker's purpose being to give the listener a correct, 
adequate, unambiguous notion of the referent. 

The emotive causes of nomination lie in the striving for a more ade- 
quate expression of one's own feelings, or a stronger effect on the hearer: 
the speaker chooses a word because it is better calculated to express or 
relieve his feelings, to make a stronger appeal to the hearer's feelings, 
or to present the referent in the light that the speaker deems desirable 
(cf. II. 15). I take the term emotive in a wide sense, as embracing also 
aesthetic motives, including stylistic purposes, and I regard as emotive, 
for instance, pet names and similar turns of speech peculiar to 
families and other small groups. The precise delimitation must be left 
to future research; we have not yet material enough to draw the line 
with exactitude.^) 

My three types correspond, with some modifications, to Wundt's 
classes (II 585): (i) Namengebung nach singiddren Assoziationen. 
Jede solche Namengebung ist gleichzeitig Wortschopfung und Be- 

^) Cf. Marty's reference to aesthetic considerations: "weil die neuartige Ver- 
wendung von Zeichen, die in gewissen Bedeutungen schon gebrauchlich sind, 
dem Vorstellungsleben grosseren Reichtum oder anderen Reiz zu geben ver- 
spricht" (Unt. 520). The aesthetic motives may be regarded as pertaining partly 
to the expressive and partly to the purposive function. 

Ogden & Richards (372 sqq.) discuss at length the distinction between what 
they term symbolic and emotive or evocative uses of language. "In strict sym- 
bolic language the emotional effects of the words whether direct or indirect are 
irrelevant to their employment. In evocative language, on the other hand, all 
the means by which attitudes, moods, desires, feelings, emotions can be verbally 
incited in an axfdience are concerned ... In symbolic speech the essential consider- 
ations are the correctness of the symbolization and the truth of the references. 
In evocative speech the essential consideration is the character of the attitude 
aroused. Symbolic statements may indeed be used as a means of evoking attitudes, 
but when this use is occurring it will be noticed that the truth or falsity of the 
statements is of no consequence provided that they are accepted by the hearer" 
{1. c. 376). 

284 GUSTAF STERN 1 1. 1 2 

deutungswandel, letzteres insofern als sich das neue Wort durcli irgend- 
eine lyaut- oder Begriffsassoziation an bereits bekannte Worter anlehnt. 
(2) Singiddre Namenubertragung. (3) Die sprachliche Metapher. 

11.13. Intentional and Unintentional. The common characteristic 
of all nominations is the fact that a selection of terms takes place; the 
speaker more or less intentionally chooses his phrase. His attention 
is directed not only towards the content of his speech, leaving the for- 
mulation of it to the automatic activities of lower centres, as in ord- 
inary discourse {die gewohnliche Sprechtdtigkeit) , but he is also thinking 
of how he is expressing himself. The speaker's intention is not only "die 
auf das augenblickliche Bediirfnis gerichtete Absicht, seine Wiinsche 
und Gedanken anderen verstandlich zu machen" (Paul, Prinz. 32), but 
his intention includes an effort to present his thoughts in a special 

The distinction between intentional and unintentional — in respect 
of the form of the expression ■ — is often very difficult to apply in prac- 
tice, but it seems necessary to make use of it. There is, obviously, an 
immense difference between the simple regular transfers, exemplified 
in ch. 12, and an elaborate poetical metaphor; but both may be trans- 
fers based on similarity. I do not see how we can distinguish the two 
theoretically except by drawing a line between intentional and unin- 
tentional. One further point, already touched upon in 7.33, may 
be mentioned. The unintentional changes, being based on psychic 
processes of a simple kind, can be classed on a psychological basis; the 
intentional changes, where the shift may be based on a purely indi- 
vidual and momentary combination, are not so classifiable. 

Generally speaking, intentional transfer (non-figurative) is not so 
difficult to distinguish from regular transfer. The group that makes 
difficulties is the third: figures of speech. There are among them in- 
dubitable instances of intentional invention, but it is no less true that 
figures of speech are characteristic of all vivid oral speech, where the 
term "intentional", in a strict sense, seems scarcely applicable. This 
circvunstance would seem to indicate a solution by means of an addi- 
tional line of demarcation, separating clearly intentional figures from 
the rest. But no one has been able to say how such a line should be 
drawn in order to be of practical use in classifying. Although a tri- 
partition, into clearly intentional, clearly unintentional, and a middle 
group which I cannot define, would perhaps be theoretically correct. 


I have, for practical reasons, to be content with a bisection of the field; 
I retain the terms intentional and unintentional, and I define them so 
as to fit the linguistic material. i) 

I take the term intentional in a wide sense. The amount of inten- 
tion directed towards each individual word may be at a minimum. 
We have to reckon with an emotional tension or enhancement (see 
II. 15) for which the speaker strives to find release, and which he 
strives to impress on the listener. He will then also strive, more or less 
automatically in respect of each separate word, but intentionally in 
respect of the whole, to keep his expression at a certain stylistic or 
emotional level, within the circle of words capable of expressing or 
carrying the intended emotional colouring, or capable of making the 
intended impression on the listener. Words arising in his mind will 
be accepted if they fit the speaker's general attitude and intentions, 
and rejected if they do not; in so far a selection really takes place 
(Cf. Carnoy 90 sqq). By taking intentional in this way, I am able 
to include under that heading the whole mass of metaphors, popular 
or poetical, which would not otherwise be possible. 

It is evident that practically any sort of sense-change that can be 
produced unintentionally, can also be produced intentionally; e. g., an 
analogical formation or a shortening leading to a sense-change. I have 
stated in ch. 7 my reasons for not making the degree of intentionality 
a basis of classification with regard to types other than the transfers.'^) 

There are clear cases of intentional, as well as of unintentional 
changes. There are also many cases in which it is difficult or imposs- 

^) Another method would be to distinguish in the first place emotive and non- 
emotive transfers; the former would be equivalent to the figures of speech, and 
would form class IV; the latter would comprise regular transfers and intentional 
transfers based on similarity, and would form class V. But intentional transfers 
based on other relations could not be moved into class VI (permutations), because 
that would destroy the unity of the class. The only way would be to let them 
form a class of their own, together with intentional naming. But that would, 
on the other hand, separate the two types of intentional transfer, which seems 
contrary to sound principles. And if we join both types of intentional transfer 
together, we are back in the classification I have adopted, with the formal differ- 
ence that I make emotive and non-emotive intentional shifts form one class, 
instead of two. 

2) Nyrop (IV 396 sqq.) summarizes the ancient dispute as to whether objects 
have their names by nature or by convention. On p. 414 sqq., he gives numerous 
French instances of intentional change. 


ible to apply the distinction; a circmnstance which attends all semantic 
classification. With regard to the doubtful instances, I propose to 
apply the following method. We know that unintentional changes 
must be constituted by psychic processes simple enough to occur 
without deliberate intervention on the part of the speaker, and invol- 
ving modifications small enough and natural enough to pass un- 
perceived by both speaker and hearer. There are a number of types of 
sense-change that can plausibly be shown to be unintentional in this 
manner. Other types are perhaps not demonstrably unintentional, but 
they are so similar to the previous group in circumstances and general 
characteristics that they can reasonably be classed along with it. All 
types which are not in this way unintentional, or closely related to 
unintentional types, are preliminarily classed as intentional. Additional 
research into the details of this question will perhaps draw the 
line differently, but at present I do not see any other way of solving 
the difficulty and evading purely arbitrary decisions.^) 

^) Wellander (Studien I 139) restricts Namengebung — corresponding in the 
main to this class — to cases where "unter Benutzung des vorhandenen Wort- 
schatzes, unabhangig von fremdsprachlichen Einfliissen, eine wirkliche Wahl 
getroffen wird". He quotes Marty (Unt. 632) for the term bewusste Uebertragung, 
and for the following characteristics: "ich meine, was bei ihr vorhanden sein 
musste, war: (i) das Bewusstsein der neuen Bedeutung in Unterschied von der 
friiheren, (2) die Absicht, sie durch das alte Bezeichnungsmittel mitzuteilen, (3) 
ein Anlass daf iir, gerade dieses Mittel zu wahlen, trotzdem dessen friihere Bedeut- 
ung nicht mit der ihm neu zugedachten verwechselt wurde" (See further Marty, 
Unt. 595 note 2, and 646). 

This definition might perhaps be accepted for intentional transfers, although 
even for them it seems to me to stress the consciousness of the process too emphat- 
ically. But for the figures of speech it would, as far as I can see, cut across the 
metaphors. In the enumeration of metaphors, I think Wellander includes 
instances that do not fall under his own definition, and in which it is very difficult 
to believe in a conscious selection. In the rapid flow of speech, where metaphors 
are created by lightning associations, there is not time to compare the primary 
meaning and the actual meaning, in three movements. A passing association is 
caught on the wing, if it agrees with the speaker's intentions, if it presents the 
referent in the desired light, and suits the style employed for the moment. If 
intentional is restricted according to Marty's definition, we should have to establish 
a third large region between the strictly intentional and the strictly unintentional 
transfers, and, as I have already stated, I am unable to see how such a tripartition 
could be effected so as to be of practical use. Wellander gives no information on 
this point. 

Sterzinger states with regard to Steigenmg (enhancement, see 11. 15), which 


11.14. Intentional or Singular? In Wundt's classification we find a type 
corresponding in part to what I have here called intentional sense-change, and 
termed by Wundt singuldrer Bedeufungswandel. Singular sense-changes are "alle 
diejenigen Brscheinungen des Wechsels der Wortbedeutungen, die aus individu- 
ellen, ah spezieUe Raum- und Zeitbedingungen gebundenen Motiven hervorgehen", 
The distinction between singular and regular is based on the fact that in the 
former case "die Ursachen die ihm bestimmen, einem in dieser Kombination nur 
einmal vorhanden gewesenen Zusammenfluss von Bedingungen ihren Ursprung 
verdanken. Zuweilen lasst sich dieser singulare Ursprung direkt dadurch nach- 
weisen, dass er auf einen bestimmten Urheber zuriickzuverfolgen ist; in vielen 
andem Fallen kann er nur aus dem ganzen Charakter des Vorgangs erschlossen 
werden" (II 584). "Singularer und regularer Bedeutungswandel unterscheiden 
sich daher im allgemeinen derart, dass jener auf die einmalige, dieser auf die 
mehrmalige unahhdngige Entstehung bestimmter Motive zuriickweist. Hieraus 
ergibt sich, dass diese Gegensatze hier einen etwas andern Inhalt besitzen als bei 
dem LautAvandel. Natiirlich ist aber auch bei dem singularen Bedeutungswandel 
nicht notwendig ein Einzelner der Urheber einer Begriffsiibertragung. Der Unter- 
schied der einmaligen und der vielfaltigen Entstehung liegt vielmehr darin, dass 
der singulare Vorgang beim Bedeutungswandel jedesmal den Charakter einer 
willkiirlichen Handlung an sich tragt, wahrend bei dem regularen die bei alien 
oder den meisten Individuen einer Gesellschaft wirksamen Assoziationsmotive 
mit dem den Triebhandlungen eigenen Zwang die Umwandlung der Begriffe 
bewirken" {II 469. Cf. Funke, Sprachform 102, Roudet 684 sqq.). 

Marty has pointed out (Unt. 594) that Wundt employs the word willkurlich in 
two senses: "einstweilen in dem Sinne, dass eine bloss zufdllige Beziehung zwischen 
der alten und neuen Bedeutung bestehe (nicht eine innere, in der Natur der Be- 
griffe liegende), andemteils in dem Sinne, dass eine solche Beziehimg eben nur 
im Zusammenhang mit Willkiir im Sinne eines WoUens und Wdhlens wirksam 
werden, dieses aber nicht Sache der Gesamtheit sondem nur einzelner sein konne". 
In the passage just quoted from Wundt II 469 it is evidently the second of these 

he considers as constitutive for the figures of speech: "Das Steigerungsphanomen 
. . . zeigt sich zunachst als instinktiver Trieb, die ProtokoUe sprechen gewohnlich 
nur in Schlagworten von seinem Auftreten, gebrauchen die Bezeichnungen "un- 
willkiirlich" und "automatisch", so elementar und plotzlich erscheint er ohne 
irgend welche vorherige Besinnung. Durch die starke Erregung aber wird er 
sehr rasch bewusst, imd seinen Aufschliessungskraft wird formlich gewaltsam 
nach alien Seiten erprobt. Ist das nun schon vielmal geschehen, so kann schliess- 
lich die sozusagen technische Seite der Erscheinung, das Vergrossern und Ver- 
vielfachen, auch absichtlich und gedachtnismassig angewendet werden" (Zs. f. 
Asth. 12, 78). What Sterzinger here describes as spontaneous and automatic is 
clearly the progressive tendency itself; what I am discussing is its expression. 
If I understand him correctly, what he calls the Aufschliessungskraft of the ten- 
dency is what I am speaking of when I say that words which agree with the pre- 
dominating tendency are adopted and made use of in preference to those that 
do not. 


interpretations that is intended, since Wundt speaks of eine willkurliche Handlung. 
That is also the sense in which Wellander takes the term. And so, if singular and 
willkurlich in this sense really coincide, there would be no difference between 
the two standpoints. 

It is regrettable, although excusable, that Wundt has not worked out his prin- 
ciples in greater detail. It seems scarcely probable that intentional and singular 
should always coincide. They may do so; for instance in the case of Mercury. 
The name of the messenger of the gods is first transferred to the swiftest of the 
planets, and later on to a liquid metal. The basis of the transfer in each case must 
have been a "singular" association in the mind of an individual, plus an act of 
intentional naming. But there are other cases where intentional and singtdar 
do not coincide. We may take as an instance the much-discussed Moneta — 
assuming that the usual theory is correct, and Moneta is a name for Juno. The 
earliest Roman mint was, according to Livy, situated in the temple or temple 
precincts of Juno Moneta on the Capitol, and in some way the mint itself received 
the name of Moneta, which was retained as the mint was later on removed to 
another place in the city. Wundt regards this as an instance of singular naming. 

The combination of circumstances leading to the shift was clearly singular. 
But was the naming necessarily intentional? Of course, we do not know anything 
about the real facts of the case, but since I am using the name merely for pur- 
poses of illustration, that is not for the moment relevant. We may assume that 
a Roman on his way up to the temple would say, "I am going up to Moneta", 
and the speaker's person or the circumstances would indicate to his interlocutor 
whether he meant the temple or the mint; just as a London financier of to-day 
may say to a friend, "I am off to Threadneedle Street", and his friend will know 
that the object of his visit is the Bank of England. For obvious reasons Thread- 
needle Street cannot become the habitual name of the Bank, but in the case of 
Moneta the shift was possible. No doubt for quite a large portion of the popu- 
lation of Rome the mint was by far the most important of the two institutions 
called Moneta, and so that meaning might become more frequent than the original 
one. This process is gradual, and in my opinion need not be at all intentional. 
The shift belongs to class VI, permutation. 

I conclude that a sense-change based on a singular combination of circum- 
stances need not necessarily be intentional; the singular situation may lead to a 
"regular" change. Since this change may be of any kind, a distinction regular/ 
singular would cut across my classification, and it is therefore ruled out. (Cf. 
Marty's criticism, Unt. 560 sqq., esp. 593 — 594). 

11.15. The Causes and Conditions of Nomination. The following 
remarks apply only to intentional transfer and figures of speech; in- 
tentional naming is of subordinate importance. 

I stated above that the cause of a nomination is the circtimstance 
that the habitual term, or the term that first offers, does not appear 
suitable for the occasion. The speaker finds that another word, which 
he is able to make use of, will serve better. It may be a more adequate 


symbol of the referent in the nionientary situation; it may communi- 
cate the relevant mental content in a more precise manner; it may 
express the speaker's thoughts or feelings with greater exactitude; or 
it may, in connection with the three factors mentioned, be more effec- 
tive in its action on the hearer, and so be better fitted for the speak- 
er's purposes. 

The principal causes of nomination are thus to be found in the speak- 
er's striving to adapt speech to his purposes; in other words, to make 
speech fulfil its functions as effectively as possible. 

With regard to the causes which lead to employing figures of speech, 
I refer to the investigations of Sterzinger, Kainz, and others. Accord- 
ing to Kainz (3 sqq.), who quotes Sterzinger, Chr. Ruths, Ch. Biihler, 
R. M. Werner, and others, our psychic processes are characterized by 
a tendency to Steigerung (enhancement; cf. Elster I 15 on die asthe- 
tische Anschauung, and Miiller-Freienfels, Psych, d. Kunst II. 27 sqq.). 
"Unter Steigerung verstehe ich jede wahrend eines geistigen Produk- 
tionsvorganges sich voUziehende Umanderung, die bewusst oder un- 
bewusst teleologisch orientiert, eine Erhohung, eine Intensivierung der 
erstrebten Wirkung zum Ziele hat. Hierher gehort: jede Verstarkung, 
Vergrosserung, Vervielfachung, Haufung (additive Vereinigung wir- 
kungsvoller Ziige), jede Kombination und Verschmelzung von Einzel- 
ziigen, die ein wirksameres Gauzes ergibt, jede Art von Konzentration 
und Zentralisation. Massgebend dabei ist nicht letzten Endes die 
mathematische (quantitative) und dynamische Erhohung, sondern die 
intensivierte Wirksamkeit, die auch gelegentlich durch Verminderung, 
Verkleinerung erzielt wird, also durch scheinbar Regressives. Das, 
worauf es ankommt, ist die Entfernung von einem aus der Gesamtein- 
stellung des Kiinstlers sich ergebenden asthetischen Indifferenzpunkt, 
nach der Seite des in seinen Augen asthetisch Wirksameren, von ihm 
als kiinstlerisch wirksamer Erkannten. Der Begriff der Steigerung ist 
ein Beziehungsbegriff, man kann ihn nur dort exakt und sicher an- 
wenden, wo eine Beziehung, eine Vergleichsmoglichkeit vorhanden ist" 
(1. c. 8). 

The human mind always remodels its material, and this remodelling 
goes in the direction of enhancement, enlargement, intensification, that 
is to say, in a progressive direction. We remember houses, or rooms, 
as bigger than they really are; when we relate an event, we emphasize 
certain traits, we bring out contrasts, and so forth, in order to heighten 

Goteb. Hogsk. Arsskr. XXXVIII: i. 


the effect: we unconsciously exaggerate when we tell a story. 
Among the psychic causes of enhancement or progression, Kainz 
first mentions the progressive substitution as a general, elementary ten- 
dency of our mental processes; this is a factor in the imaginative activ- 
ity. Of the specific psychic causes, the first and most important is 
the need of "erhohte Abreaktionsmoglichkeit", increased possibilities 
of emotional relief. We seek intensified expressions for that which 
agitates our feelings, and this leads to progressive substitution, especi- 
ally in the case of an intense excitement with a polemical or apologetic 
purpose. In addition to the pleasure in spiritual productivity, we have 
to reckon with "das kathartische Moment der Aussprache, die bef reiende 
Entladung die bei steigernder Aussprache wirksamer ist als beim tat- 
sachen-treuen Bericht. Die Steigerung kommt hier den Ausdrucks- 
bediirfnis entgegen. Das fallt in die kundgebende Funktion der 
Sprache". A second motive lies in the need of and pleasure in mental 
productivity, the formative tendency of imagination, "der lyust am 
fabulistischen Trieb", the desire to be artistically active and creative. 
The speaker feels that he is the master of his material, and is not fet- 
tered by it. 

A third motive is the elementary tendency towards what is fresh, 
important, and most progressive in any direction or of any kind; the 
desire for expansion, for an extension of the ego, for possibilities greater 
than those offered by real life. Here the emotional factor is pre- 
dominant. A fourth motive is the desire for effect, the striving to give 
to an utterance a higher tension, a more intense impressivity. We 
underline, we emphasize contrasts; we seek the effective, the palpable. 
(Kainz 26 sqq.). 

With regard to the causes of our finding pleasure in the poetic figure, 
Sterzinger (Arch. 29) states that the most important is the fusion 
of two notions or images, and, in the second place, the occur- 
rence of emotional elements and pseudoperceptions (centrally evoked 
perceptions) . 

On the whole we find, as we might expect, that the factors 
leading to the use of figures of speech involve the expressive 
and effective (purposive) functions of speech, not the symbolic and 
communicative functions. To the expressive function belong the first 
three of those enumerated by Kainz. The problem seems to require 
much additional systematic research before it can be said to be 


satisfactorily elucidated, and I have to leave it with these brief 

With regard to the conditions of nomination, we cannot make precise 
statements. We can only say that the possibility of nominations is 
limited by the hearer's ability to understand the new meaning; in other 
words, the fuller the context and the more clearly the referent is indi- 
cated through the context, the greater liberty the speaker enjoys in 
his choice of terms. When Keats begins an Ode, "Thou still unravished 
bride of quietness. Thou foster-child of silence and slow time", these 
far-fetched metaphors are comprehensible to the reader, and thus 
possible, because the heading of the poem gives its topic: Ode on a Gre- 
cian Urn. 

The speaker's choice is generally not so wide. The intentional na- 
ture of the transfer or figure of speech opens a much larger field to the 
speaker's imagination than do the unintentional psychic processes, 
but this does not mean that the process is entirely arbitrary. The 
main groups show certain predominant types. We use personal names 
for what a person produces, place names for what happens at or comes 
from a place, and so on; and new instances are easily formed and as 
easily understood in analogy with the great mass of shifts with which 
every speaker is familiar. 

11.2. Intentional Naming. 

Intentional naming is the coining of a new name to denote a referent, 
whether new or old. I shall mention briefly some of the more important 

Entirely new words, at least successful ones, are comparatively rare, 
The most famous of them is probably gas, coined about 1600 by B. van 
Helmont, who states that "ideo paradoxi licentia in nominis egestate 
halitum ilium 'gas' vocavi, non longe a 'chao' vetere secretum" (quoted 
Oertel 163). In modern business such creations are somewhat more 
common: kodak, Tono-Bungay. 

In most cases, old words or parts of them are employed to form new 
words. The use of shortenings has been mentioned in the preceding 

^) Cf. Wundt II 586 sqq., Marty, Unt. 662, Carnoy 254 sqq., Wellander, Studien 
I 138 sqq. (but note that some of Wellander 's instances are not true cases of 
nomination), Beckman 150 sqq., 167. 

292 GUSTAF STERX 1 1. 2 

chapter. Portmanteau- words are formed by compounding parts of 
old words: anecdotage, galumph, slithy (see Pound and Bergstrom). 

Naming on a large scale takes place in connection with notable ad- 
vances in knowledge, or other modifications or upheavals in our social 
or intellectual life. The Great War is an outstanding instance. Accord- 
ing to Nyrop (Etudes 29) it manifests its influence on language mainly 
in four different ways, the creation of new words, the giving of new 
meanings to old words {i. e., intentional transfer), invasion of slang 
words in literary language, and extensive borrowing of words. With 
regard to the first class, Nyrop states, however, that he has found 
surprisingly few really new words. I may add that many of Nyrop 's 
instances are analogical formations, thus amerrir (an aeronautic word 
on the analogy of atterrir) desannexion , ententiste, Kaiseriole, a humor- 
ous formation after Carmagnole, and others. 

The puristic tendency in Germany has led to the coining of numerous 
words intended to replace technical terms of the international type: 
Fernsprecher, Radfahrer, Drahthericht, Kraftwagen, Selhstfahrer, Schrift- 
leitung, Anschrift, Seitenweiser, Beistrich, and many others. English 
instances of the same kind are foreword, folklore, hirdlore, bodeftd, better- 
ment, forebear, belittle, wheelman, happenings, etc. (Cf. Fowler, MEU 
s. V. Saxonisms, and 9.331 note). Intentional naming and analogy here 
touch. The same type of compounds is often used to obtain an emotio- 
nal effect, generally humourous: devil-dodger or sky-pilot for clergyman, 
inkslinger for writer, sawbones for surgeon, and so on. Trotters for 
feet, peepers for eyes, shiners for precious stones, and other slang words 
of similar type, belong here. 

The compound type is the one in which a sense-change might be 
said to be involved, but even there the matter is somewhat doubtful, 
and I mention it only for the sake of completeness. When a vessel 
was built capable of sailing in the air, it was termed a ship; this is a 
substitution, as explained in ch. 8. But if we assume that the vessel 
is at first called an air-ship, it might be said that since this is a totally 
new compound, we have no longer a case of substitution, but of inten- 
tional naming. However, by being called air-ship the new referent is 
classed as a ship, and the range of the word ship is thus increased, 
since to linguistic feeling an air -ship must be a species of the genus 
ship. In this way it might be argued that the naming indirectly causes 
a sense-change. If, subsequently, the simplex ship is used to denote 


the air-ship, there are alternative explanations. Wellander regards 
it as a shortening; it niay also be a case of genus pro specie. Cf. 10.412. 
Substitution and intentional naming here come very close to each 
other; either may be adduced in explanation. Cf. 8.2. 

11.3. Intentional Transfer (Non-figurative). 

11.31. General Remarks. Intentional transfer is the intentional 
denoting of one referent by a name that really belongs to another, 
the reason being a desire for a more adequate symbolization or com- 
munication; that is to say, the causes are cognitive, not emotive, in 
character. The transfer is founded on some similarity or other relation 
between the primary referent and the actual referent. i) (Cf. Wundt 
II 589 sqq.) 

Intentional transfer is distinguished from regular transfer by being 
intentional (see 11.13). Figures of speech differ from intentional 
transfer by the emotional nature of their causes (see 11. 4). 

Some writers have sought to obviate the difficulties involved in the 
two distinctions just mentioned by extending the term metaphor to 
cover the whole domain of transfers and figures. But this only shifts 
the difficulty to another place in the system, We cannot get rid of it, 
since the two distinctions are too important to be neglected in any 
system of classification. 

With regard to the further classification of intentional transfers, 
the question is whether there are any psychological factors involved 
that offer a suitable basis for distinctions, or if other points of view 
should be applied. Since this type is closely related, on the one side to 
metaphors, on the other side to regular transfers and permutations, 
it is desirable to find principles capable of application to all these 

It seems to me that there are two possibilities. The first of them is 
to distinguish transfers based on similarity from those based on other 
relations between the two referents involved. We shall find that the 
same distinction is applicable to metaphors, and that it constitutes one 
difference between regular transfers and permutations. Since it is 
thus evidently an important point of view, I shall take it for my first 

1) The primary referent is the original referent of the word used. The actual 
referent is the one to which the word is actually applied when transferred. Simi- 
larly I use the terms primary and actual meaning. 


basis of classification. Secondly, it is possible to class the transfers 
according to the parts of speech involved. It is not quite clear at 
present whether there is really a constitutive difference between the 
sense-changes in nouns, adjectives, verbs, and particles, but the distinc- 
tion between these classes of words is well-known and offers a handy 
method of reducing the size of the groups. It may therefore be em- 
ployed with advantage until a better method is found. 

11.32. Transfers based on Similarity. Intentional transfers on a 
large scale occur in the naming of new referents in the course of scien- 
tific, technical, or social progress. 

English instances are to be found in most scientific works that handle 
new material or new problems. In the present treatise, the terms 
substitution, nomination, permutation, are instances of intentional 
transfer based on similarity; the similarity is sometimes admittedly 
vague, but it is there nevertheless. As in regular transfers, the simil- 
arity may apply to the appearance, the fanction, the position, etc. 
of the referents. We speak of the walls of a cell, and the vertebra 
supporting the skull is named Atlas. Bain (I 163) instances from ana- 
tomy pons varolii, the labyrinth of the ear, etc. "All terms derived from 
mechanical forces are adopted in the description of social forces: im- 
pulses, propulsion, momentum, resistance, inertia, cohesion, attraction. . . 
Again, societies are described as rude, polished, advanced, complicated" 
(Bain I 142). Most of Bain's instances of "intellectual similitudes", 
being taken from poetry or elevated prose, have a tinge of the emotive 
element, and thus would belong to metaphors rather than to the present 
type (cf. Wundt II 588).!) 

In popular speech, it is often very difficult to ascertain if a transfer 
is intentional or regular: Engl, crane, German and Swedish Kran, for 
a hoisting apparatus, probably from the likeness of the long beam and 
hanging rope to the neck and bill of the bird; Engl, saw-horse, German 
Bock; wandering, used of diseases, pains, etc., 'moving from one part 
of the body to another (without clearly ascertained reason)'; and many 
others, may be due to an act of intentional naming, but unintentional 
origin is also possible. 

^) Some of Wundt's (II 591) and Wellander's (Studien I 139 sqq.) instances 
are not real transfers, but should rather be interpreted as exemplifying a fresh 
delimitation of the referent: substitution. 


11.33. Transfers based on Other Relations. 

11.331. Proper Names for Objects. Proper names are employed 
to denote scientific units of measurement, the transfer being intentional 
and in honour of the possessor of the name: ohm, ampere, volt, coulomb, 
brinell (cf. Weekley, Words 40). 

In terms of popular formation, the development is often uncertain. 
According to Greenough & Kittredge (256) nobody ever said sandwich 
lunch and spencer coat, the short terms sandwich and spencer being used 
from the beginning (Weekley, Words 39 — 40. The names are from an Earl 
of Sandwich and an Earl of Spencer). In other cases where a proper 
name is used for an invention or discovery, shortening is demonstrable. 
Thus a shrapnel is a shortening of the full name shrapnel shell, named 
from the inventor, General Shrapnel (cf. 10.431). This may be the case 
also for galling, hotchkiss, lebel, mauser, remington, Winchester, chassepot, 
mackintosh, burberry, bluchers, Wellingtons, chesterfield, raglan, havelock, 
Vandyke, broiigham, macadam, and others. (Cf . the lists given by Fischer 
Engl. Stud. 23, Aronstein ib. 24, Reinius 148 sqq., Ostberg 98 sqq., 
Nyrop IV 376). 

Other cases, outwardly similar, may be permutations (cf. 13.24). 
It is necessary, if we want to be certain, to trace the history of each 
word. The NED states that, for instance, tam 0' shanter is a shortening 
of tam 0' shanter bonnet, but the earliest quotations show the shortened 
form. What evidence is there for a shortening? 

11.332. Place-names for Products or Events. The use of place- 
names for products of the place can no doubt be explained in more 
than one way, and it must in many cases remain doubtful what the 
correct explanation is. When we speak of mokka, Java, madras, calico, 
china, a basque, a jersey, camembert, holland, cremona, bordeaux, cham- 
pagne, boston (a game at cards), etc., the names in their secondary use 
may be shortenings of mokka coffee etc. (cf. 10.431, and see Greenough 
& Kittredge 255, Weekley, Words 47; Nyrop IV 391). Similarly 
when an angora is used for an angora cat. (See Efvergren, passim). 

When place-names are used for events which have happened at the 
place, intentional transfer is also possible, but I believe it is less likely. 
An alternative explanation is permutation (cf. 13.251); but Waterloo, 
for the battle of Waterloo, may be due to stylistic considerations, in 
which case we have a metaphor (11.77). 

296 GUSTAF STERN 11-333 

11.333. Christian Names. This is, I think, the best place to class 
the giving of a christian name to a newborn individual: it is undoubt- 
edly an intentional naming, and even if the choice is not entirely 
uninfluenced by emotional factors, we can scarcel}^ class the name as 
a figure of speech. 

Proper names have characteristics of their own, which are not alto- 
gether easy to disentangle, and which have occasioned much debate. 
It is not necessary to go into the matter at length. I shall only state 
that Y^rious difficulties seem to be avoided if we regard Richard, as 
denoting Mr. Richard Brown, and Richard, as denoting Mr. Richard 
Smith, as a pair of hononyms. The Richards do not form any distinct 
category of htunan beings, and we are therefore not justified in speaking 
of the name as being still the same name, although with different mean- 
ing, as we do when, for instance, ship is used to denote various new 
inventions in the way of vessels: these belong to a recognizable category 
with peculiar characteristics of its own. (Cf. Conrad, Arch, ig, 399). 
This interpretation of the facts is in strict accordance with my defini- 
tion of meaning, which makes meaning dependent on, among other 
things, the actual characteristics of the referent. 

Possibly we should explain in the same way the application of a 
family name to a new member, even if it is not intentionally transferred, 
in the strict sense of that term, but inherited or acquired by marriage 
according to the customs, regulations, statutes, etc., of the community. 

11.4. Figures of Speech. 
11.41. Introductory and Definition. Figures of speech are inten- 
tional transfers which involve emotional (incl. aesthetic) factors. They 
are intended to serve the expressive and purposive functions of speech 
better than the "plain statement".^) 

^) Bain (I 135) takes into consideration only the latter point: "A figure of 
speech is a deviation from the plain and ordinary way of speaking for the sake 
of greater effect". Note that the deviation as such arrests the attention of the 
hearer, and thus is of importance for the effect of the figure. "Surprise tends to 
increase the intensity of every emotion with which it blends, or by which it is 
rapidly followed", and "the effect of surprise is to make us attend to the event 
that surprises us" (Shand 422, 430). The same observation is thus formulated 
by Selz: "Stimmt die iiberwiegende Mehrzahl der Bestandstiicke eines Erlebnis- 
komplexes in irgend einer Hinsicht iiberein, so hat ein demselben Komplex ange- 
horendes Bestandstiick, das in dieser Hinsicht von ihm abweicht, die Tendenz, 


The distinction I have adopted, between purely cognitive processes, 
and those that are more or less emotive, is of course very difficult to 
apply in many individual cases. Plain statements and figures merge. 
The emotional element may also fade in the course of time, so that 
the figurative expression assumes the appearance of an ordinary 
name for the referent (see 14.52, adequation). In that state figures 
are indistinguishable from other transfers or permutations, if we do 
not know their earlier history. 

When we approach the question of classifying the figures of speech, 
the problem arises whether the traditional classification of the rhetori- 
cians should be retained, or other principles adopted (cf. Wundt II 
594 sqq.). The old classification seems to have no uniform basis. It is 
partly based on logical considerations, partly on the form of the expres- 
sion, partly on the nature of the subjective attitude expressed in the 
figure. It might be possible to substitute a classification based alto- 
gether on psychological factors. But here we encounter two difficulties. 
It is evident from the analysis that different types of mental content 
can be expressed through one and the same figure (cf. Sterzinger, Arch. 
37> 365, 383)- Secondly, no satisfactory psychological investigation 
of the figures of speech has as yet been published. Fortunately, it is 
not necessary for the purposes of the present work to go into the matter 
at length. Many figures of speech do not lead to permanent sense- 
changes of single words, and are thus of no further interest to us. The 
only figures we have to discuss are metaphor, hyperbole, litotes, irony, 
and euphemism, and it does not appear impossible to establish a working 
hypothesis explaining their nature and furnishing reasonable defini- 

beachtet zu werden" (Selz II 496). In accordance with this view, I regard as 
figurative, for instance, the use of spurious conditional clauses to express an 
antithesis: if something was done, it was he that did it. The antithesis as such is 
not intensified, but the listener's attention is called to it by its being expressed in 
an unusual way. It would be wrong to analyse this form on the analogy of 
conditional clauses used for imperatives; in that case the conditional clause is a 
down-toner. — I refer to Bain I 135 note for the import of the term trope, which 
I do not use. See also on these matters Elster I 40 — 41, 374 sqq., Meyer, Stil- 
gesetz no sqq., and Larsson 61, 63. 

^) Synecdoche and metonymy are often regarded in semasiological works as 
main classes of sense-change, but from the point of view which I have adopted, 
they must be split up into groups belonging to different classes. Synecdoche, 
according to Bain (I 182 sqq.) is founded on similarity, and consists in putting 

298 GUSTAF STERN 1 1. 42 

11.42. The Delimitation of Metaphors. The true definition of 
metaphors has been the subject of much controversy (see Sterzinger, 
Arch. 37, 363 sqq., with quotations from various authors, and Wundt 
II 594 sqq.). The previous definition of figures of speech provides a 
genus proximum; we have now to find a differentia specifica serving 
to discriminate metaphors from other figures. 

Stahlin (Met. 14) gives a definition that may be rendered as follows: 
"Metaphors are figures of speech in which a referent is designated by 
the name of another referent in such a fashion, that (i) the transfer does 
not involve an essential identity of the two referents, (2) the designa- 
tion is taken from another sphere of experience than that to which the 
actual referent belongs, and (3) the process of transfer is not ex- 

The first criterion mentioned by Stahlin — that the transfer does 
not involve an essential identity of the primary referent and the actual 
referent — serves to distinguish metaphors from regular transfer, in 
which the point of similarity is some essential characteristic common 
to the referents. I take this fact as connected with the distinction 
between intentional and unintentional, assuming that an essential 
common element makes it possible for a word to be transferred unin- 
tentionally from the primary referent to the actual referent, the ensuing 
modification of meaning passing unperceived. On the other hand, 
when the common element is less prominent (see 11.53) it is perceived 

(i) the species for the genus, (2) an individual for a genus, (3) the genus for the 
species, (4) the Concrete for the Abstract, (5) the Abstract for the Concrete. In 
my classification, changes of these kinds are classed as intentional transfers, 
metaphors, regular transfers, or permutations, according to the circumstances of 
the change. 

Metonymy (Bain I 186) is founded on contiguity, and consists in naming a thing 
by some accompaniment. The accompaniment may be (i) the sign or Symbol, 
or any significant adjunct, (2) the Instrument for the Agent, (3) the Container 
for the thing contained, (4) an Effect for a Cause, (5) a Maker for his Works, 
(6) the name of a Passion for the name of its Object. Closely related to these forms 
of metonymy are the forms of synecdoche founded on contiguity: (i) Naming a 
thing by some Part, (2) the reverse operation of using the Whole for the Part. 
Like the synecdoches mentioned above, these forms split up into various classes 
in my system. 

1) Bain (I 159) says that a metaphor "is a comparison implied in the mere use 
of a term". Cf. Carnoy 275. On the origin of the metaphor, see H. Werner, Ster- 
zinger, and Cassirer, Spr. u. M. 68 sqq. 


and apprehended only in circumstances which justify us in calling the 
transfer intentional. Essential is a vague term, but I cannot find a 
better one. It must be taken to mean essential for the momentary 

If intentionality and the lack of essential identity thus are parallel 
phenomena, it would seem as if we might content ourselves with one of 
them as a criterion for metaphors. However, in my opinion, metaphors 
are not always founded on similarity, but sometimes on other relations 
(see 11.55), ^^^ fo^ these latter there can of course be no question of 
identity, essential or not. The criterion of intentionality is therefore 
indispensable. I retain the other criterion as a useful addition, applic- 
able to the large group of metaphors founded on similarity, and serving 
to discriminate them from regular transfer. 

Stahlin's second criterion — that the designation is taken from 
another sphere of experience — is evidently closely connected with the 
one already discussed, but, as far as I can see, they do not agree. 
"Sphere of experience" is, admittedly, a vague term; essential identity 
is somewhat more definite, and should for that reason be given the 
preference. When we use the word jew for a' grasping money-lender, 
there is no essential identity involved, since the money-lender is not 
literally a jew; the designation is figurative. But it seems to me doubtful 
whether we should say that the designation is taken from an alien 
sphere. I think it best, therefore, to reject this criterion. We may 
adopt, for descriptive purposes, Wundt's formulation concerning the 
metaphor as "eine Gesamtvorstellrung ... in der disparate Telle gemischt 
sind" (Wundt II 600), taking disparate to mean not identical in any 
essential respect. 

Stahlin's third criterion — that the process of transfer is not ex- 
pressed — is equivalent to saying that the relation between the prim- 
ary referent and the actual referent is not indicated, but has to be 
educed by the hearer. This serves to discriminate metaphors from 
similes, comparisons, and allegories. (Cf. 11. 51 and 11.52, and Bain's 
definition of the simile: the simile consists in the formal and avowed 
comparison of one thing to another (I 170). See also Stahlin, Met. 95 
sqq., and Sterzinger, Arch. 37, 384). 

We have thus far been able to establish criteria distinguishing meta- 
phors from regular transfers, on the one hand, and from similes, com- 
parisons, and allegories, on the other hand. There remains the diffi- 

300 GUSTAF STERN 1 1. 42 

cult problem of discriminating metaphors from the other four figures 
entmierated above. The difficulty lies especially in the fact that these 
figures are so often combined with metaphors. The hyperbolic word 
in / have oceans of time, is also metaphorical. I shall follow what I 
believe is the usual practice in classing such instances as hyperboles, 
noting that they are at the same time metaphorical. The first step is 
to set aside, for the moment, euphemism, which is intended to tone 
down, while the other figures are intended to enhance (11.15); and 
irony, which has peculiar properties of its own (see 11. 8 and 11. 9). 

It was pointed out above that the enhancement causing a figure of 
speech might take different forms, and I suggest that we may deter- 
mine the delimitation of the figures on this basis. We might say that 
a metaphor expresses an enhancement taking the form of a fusion of 
two disparate notions, referring to referents not essentially identical; 
a hyperbole expresses an enhancement taking the form of an enlarge- 
ment or multiplication; and a litotes expresses an enhancement taking 
the form of a reduction or diminution, that is to say, a contrast of a 
special type. 

We can then define a metaphor as a figure of speech in which (i) the 
enhancement is the result of a fusion of two disparate notions, i. e., 
there is no essential identity between the two referents involved; and 
(2) the relation between the two referents is not expressed. 

With regard to the four other figures, euphemism is, as stated, intend- 
ed to tone down, not to enhance, while irony involves no immediate sense- 
change. These two figures have therefore to be discussed separately. 

In the two remaining figures, hyperbole and litotes, a sense-change 
(a transfer) occurs only if the figure is at the same time metaphorical, 
and the shift is due to the metaphor, not to the hyperbole or the litotes. 
I have therefore not considered it necessary for my purposes to distin- 
guish in the following lists of instances, metaphors that are also hyper- 
boles or litotes, from those that are not. 

This different treatment of the figures may seem inconsistent, and is 
perhaps so from an outward point of view, but it seems to me to be 
expedient in the extant circumstances, considering the scantiness of 
really reliable material. And I submit that the theoretical investiga- 
tion of the figures of speech has not arrived at such definite results that 
we are able to state with confidence what is theoretically consistent and 
what is not. (Cf . Sterzinger's statement on the psychological correlate 


of the figures, quoted 11. 51). I therefore consider myself at liberty 
to arrange the material with regard to its semantic properties, as indi- 
cated above. These are, after all, my chief objective. 

In accordance with these principles I now proceed to a theoretical 
discussion of metaphors, hyperboles and litotes, followed by lists of 
instances, in which these three figures are not separated. The 
chapter is concluded by paragraphs on irony and euphemism. 

11.5. Metaphor (incl. Hyperbole and Litotes). Theoretical Discussion. 

11.51. The Psychic Process. The Speaker. Some writers on meta- 
phors have unaccountably neglected the productive aspect, and con- 
centrated on the analysis of the hearer's mental processes. It is evid- 
ent, however, that the genesis of a metaphor lies entirely with the 
speaker. It is the speaker who perceives the association on which the 
metaphor is based and formulates its peculiar linguistic expression. 
The characteristics of metaphors must therefore be sought in the 
speaker's mental processes and in the linguistic form in which he gives 
them expression. The hearer's part is to adapt himself to the speaker's 
mood and to comprehend the unusual S3^mbol. 

I shall take the simple instance of the word lion being used to denote 
a brave man. We have to imagine this to happen for the first time — 
it is the coining of metaphors that is being discussed, not the repetition 
of habitual metaphorical expressions. The speaker will perceive a 
point of similarity between his referent (a brave man) and a lion: the 
quality of courage. He does not make a comparison, he simply employs 
the word lion to denote the brave man. It is evident that such a pro- 
cedure would be impossible if it did not rest on a traditional practice. 
We are accustomed to this method of using words, and we comprehend 
it even if the point of similarity should be somewhat vague or far-fetched. 

The matter may be illustrated by a diagram. 






'a brave man' ^^ 

^^ 'lion' 

(Speaker's apprehension 



of the referent) 


The man 

The lion 

ah c 

c^ d e 

302 GUSTAF STERN 11-51 

In Referent I, the man, only three main characteristics (Merkmale) 
are designated, in order not to complicate matters unnecessarily. Of 
these c is assumed to be the quality of courage. In the lion (Referent 
II) there are also three characteristics, of which courage is one, but 
since it appears in another aspect in a lion, I have given it the symbol 
Ci- In a corresponding manner, the meanings, that'i?; to say, the subjec- 
tive apprehension of the referents, are given each three main elements, 
of which y and y^, respectively, correspond to the qualities of courage 
in man and beast. 

For the speaker, the mental process starts from Referent I, the man. 
The speaker is talking of one of his main characteristics, in the extant 
circumstances, the quality of courage, and desires to denote this in 
some manner more striking than the use of an adjective of suitable 
meaning. The quality c, apprehended as y, occurs also in the speak- 
er's apprehension of a lion, as yi, and according to the psychological 
law that not only identical, but also similar elements may serve as 
stimuli for the reproduction of a total, y is also capable of evoking 
the complex II, word and meaning, the latter = the apprehension of 
the referent. But the speaker's purpose for the moment is to find a 
word. The whole complex is therefore not actualized, but only the part 
of it that satisfies the task {die Aufgahe) and falls in with the deter- 
mining tendencies; that is primarily the word, lion. According to the 
law that if an element is associated with a complex entity, it is asso- 
ciated with each element of the complex, it is possible for the idea of 
courage, y, to call up directly the word lion, without necessarily evoking 
the whole complex II (cf. 4.24). Irrelevant elements of meaning — 
as the yellow colour, the nocturnal habits, the tail, etc. — are inhibited; 
they would be merely disturbing (cf. 11.54). "^^^ process is indicated 
by the arrow in the diagram. 

I refer to the following argumentation in Ebbinghaus-Biihler (680): "Auch 
wenn gegenwartige Empfindungen oder Vorstellungen mit friiher dagewesenen 
nicht identisch, sondern ihnen nur dhnlich sind, rufen sie Vorstellungen der mit 
jenem Ahnlichen friiher verbunden gewesenen Gebilde hervor. 1st z. B. urspriing- 
lich eine Gruppe von Gebilden ab c d e erlebt worden, so wird die Wiederkehr 
der Glieder a und b die den iibrigen Gliedern entsprechenden Vorstellungen yds 
wachrufen; aber dies wird auch schon geschehen, wenn nicht a b, sondern die 
ihnen nur ahnlichen Eindriicke a^ b^ jetzt in der Seele durch die ihnen ent- 
sprechenden Ursachen erzeugt werden. Um so leichter und sicherer begreiflicher- 


weise, je grosser, und um so schwerer und seltener, je geringer die zwischen a b 
und a^ fci bestehende Ahnlichkeit ist". 

In criticizing the traditional theory of association b}' similarity, Ebbinghaus- 
Biihler argue as follows, which has a direct application to the matter in hand. 
The process does not go from the impressions ab to the similar impressions a^ b, . 
But ii a b have previously been connected with c d, the process goes on to c d, 
not only from a b, which have previously been connected with c d, but also 
directly from a^ b^ to c d, or rather to the thought of these impressions, yd. 
If a child cries papa to a stranger, it is unlikely that the stranger first evokes 
in the child the idea of the father, and so the word. Rather the total impression, 
which is similar to that of the father, leads on directly to the word. There is 
nowhere any reason for the impression a 6 to transfoi;pi itself into the similar 
impression a^ b^. But according to our general notions of associative processes 
it is practically a necessary assumption that not only a b but also a^ b^ should 
have the power directly to call up c d (1. c. 687). 

Mental structures may resemble each other in such a manner that they have 
a smaller or greater number of isolable common elements, as may be represented 
by the symbols abed and c d m n. If the group abed is given, the elements 
c d contained in it may in favourable circumstances evoke the ideas fx v We 
have then in immediate succession the two experiences abed and c d jxv, but 
again not owing to any special power of reproduction inherent in similarity, but 
owing to the effect of the general law of association in these particular circum- 
stances. This is the case with all analysable similarities (1. c. 688). 

The theory favoured by some writers (see for instance the quotation 
from Marty, 11. 13 note) that the metaphor is based on a comparison, 
would involve, applied to my diagram, the process running on from 
y to 7i, and thence to the word lion. Such a process is what one would 
expect as the psychic correlate of a simile, "a formal and avowed com- 
parison of one thing to another" (Bain, quoted 11.42 note), in which 
two similar elements in two referents are simultaneously in mind. In 
view of the differences between metaphors and similes, not only in the 
verbal formulation, it does not seem probable that they should express 
the same manner of apprehending, even if their referents are identical. 
However, the matter is controversial, and in cases where speech is 
slow, and time is afforded to set the primary and secondary referents 
side by side for purposes of comparison, possibly the same verbal form 
may cover a different process. Individual variations are also probable, 
but on the whole it seems to me that the explanation given here is 
likely to be valid for the majority of metaphors in speech.^) 

^) Sterzinger's analysis of a typical case seems to me to corroborate this view. 
He states that he likes to employ, personally, metaphorical expressions, but when 


In SO far as the intentional selection of words is for the purpose of 
relieving the speaker's feelings or of making a special impression on the 
listener, the speaker's and the listener's relations to the metaphor must 
be different. The speaker, as it were, sees the metaphor from within, 
from the creator's point of view, though we niust not suppose that 
this is always, or even in most cases, a fully conscious attitude on his 

In so far as the metaphor presents the referent in a certain light, or, 
as Stahlin expresses it, fuses the apprehension of the actual referent 
with elements from another sphere, the speaker's and the listener's 
relations to it are much the same. For the speaker, the notion must 
logically be primary, the word secondary, but it is not known at pre- 
sent if this sequence is always a fact. 

11.52. The Psychic Process. The Hearer. The comprehension of a 
metaphor is not essentially different from the comprehension of any 
other phrase, as analysed in ch. 6. It is always the context that guides 
the hearer to a rapid and correct actualization of that particular ele- 
ment of meaning which is intended by the speaker. 

Returning to our instance {lion) we find that, theoretically, the pro- 
cess for the hearer begins with the word. In reality, however, the 
hearer knows, in all normal cases, what the topic of conversation is, 
and in a large majority of cases he will also be acquainted with the 
speaker's opinion of the third person who is the topic. The word lion 
is therefore immediately referred to the referent 'brave man' (Referent 
I, see the diagram), to which, however, the word is not unconditionally 
applicable. In this case, as in most other cases, it is one element of 

he is writing for a newspaper he changes afterwards to the easier form of a simile. 
He has developed the habit of first formulating his thoughts in the form that is 
congenial to him, and then of making a kind of translation into the speech of 
his less scholarly readers, with a more frequent use of comparisons, similes, "it is as 
if", and other stylistic tricks, instead of the concentrated metaphorical expres- 
sion. On the reason for the different effect of these forms he says: "Im Vergleich 
und im Gleichnis sind Bild und Sache weit und weiter getrennt, man hat Zeit 
fiir die psychischen Operationen und braucht sie daher auch nicht in dieser In- 
tensitat vorzunehmen; bei der Metapher stehen sie raumlich und zeitlich enge 
beisammen: die geistige Arbeit muss rascher und intensiver erfolgen: Tacitus und 
Horaz sind eben schwerer zu lesen als Livius und Ovid". (Arch. 37, 384 — 
385). But if similes and metaphors thus call up different processes in the hearer's 
mind, is it probable that the process in the speaker's mind is identical? (Cf. 
the quotation from Stahlin, 11.54). 


the many in the meaning of lion that is intended by the speaker. Only 
this element is capable of combining with the hearer's apprehension 
of Referent I (the man). Or, turning the matter round: lion is a mo- 
tive of reproduction for all the elements of its own meaning; the refer- 
ent 'man' can be a motive of reproduction only for the elements applic- 
able to it. The two motives converge on that element ('courage'), giv- 
ing it a predominant position, and evoking it to the exclusion of the 
other, more or less irrelevant, elements. (Cf. 6.32 with quotations from 
Poppelreuter, Selz, and others). We have here a specialization of 
meaning. The remaining elements of meaning of lion, perhaps especi- 
ally the emotive elements or an awareness of the sphere to which the 
primary referent belongs, are evoked more or less, forming a back- 
ground or fringe to the actual referent (cf. on this point 11.54). 

The degree of "preparation" necessary for the comprehension of a 
metaphor varies with the degree of similarity or intimacy of relation 
between primary and secondary meaning. Even if it is of the most 
elusive kind, the hearer will make a correct reference, he will under- 
stand it correctly, en fonction du schema d'ensemble, but a correspond- 
ingly definite preparation is then required. Thus the metaphors in 
the first lines of Keats' Ode on a Grecian Urn would be incomprehens- 
ible if the heading had not previously informed the reader what refer- 
ent the phrases were intended to designate. 

It must again be emphasized that here, as in all ordinary speech, the 
basis for the comprehension of each single word is not that word alone, 
but also a mass of context: concomitant circmnstances, situation, 
knowledge of the topic, of the speaker's habits and opinions, etc. (cf. 


A difference between a metaphor and a simile or a comparison is 

the fact that in the former case the nature of the relation between 

primary and actual meaning is not expressed (cf. above 11.42 and 

11.51). The hearer has to deduce for himself from the context that 

it is a similarity between the man and the lion that is the common 

element in the instance just quoted. The hearer thus has to create 

for himself the relation between the two referents, and has to take a 

more active part in the process of comprehension than in the case of 

a simile or a comparison. 

11.53. The Point of Similarity. An important factor in metaphors 
"based on similarity is the point of similarity, the common element, 

Goteb. Hogsk. Arsskr. XXXV III: i. 20 

306 GUSTAF STERN 11-53 

often called the tertium comparationis. Earlier writers have not much 
to say on it: it must be present to the mind, it must be easily found, 
clear, striking, and adequate. 

Sterzinger has investigated the problem, and his reports show that 
in most cases a mediating item is present to the speaker's mind. This 
item, the point of similarity, may be constituted by imagery or feel- 
ings, but is mostly constituted by a configuration (Gestalt). The con- 
figuration xnay be based on elements of the same sensory quality {Sin- 
nesgehiet), or on elements of different sensory qualities. Sterzinger 
suggests that the purpose of a poetic figure might be a repeated pro- 
duction of the configuration, first in the referent, then in the illustra- 
tive figure, perhaps in order to make it more easily comprehensible to 
the listener. The speaker (the poet) would have to perceive the con- 
figuration in the referent, and then find it again in another referent. 

Sterzinger further states (Arch. 37, 396 sqq.) that the aesthetic asso- 
ciation, the fact that an element of one notion adduces another notion 
which contains a similar element, is subconscious (or unconscious), and 
that it does not often become subsequently conscious, or at least it 
does not pass the lowest stages of consciousness. This would explain 
why the statements of rhetoricians and aestheticians concerning the 
common element are so meagre. Sterzinger leaves it an open question 
whether this circmnstance is conditioned by the unimportance of the 
element, or if it is a consequence of its psychological nature. 

I think that the lack of obtrusiveness of the common element is 
explicable in the following way. The aim of the speaker employing a 
metaphor is two-fold: (i) to designate the referent; in this respect the 
metaphor is on a line with all other expressions used in ordinary speech; 
(2) to invest the referent with a certain colour, certain associations 
carried along with the word from its primary meaning and its sphere. 
The common element must be prominent enough to permit of the 
metaphor being applied without difficulty to the actual referent; being 
assisted, as previously explained, by context. Once this stage has 
been attained, the common element generally ceases to be of 
interest. The actual referent itself is present to the mind, the aware- 
ness of the primary referent is present in so far as it is necessary for the 
right comprehension of the metaphor; and for the full comprehension 
to develop the listener relies chiefly on the whole import of the word 
and of the two referents, not on any one element of meaning of the 


word. The hearer must himself educe the relations between actual and 
primary referent if they are to affect him. 

The part played by feelings as common elements is emphasized by 
Miiller-Freienfels (Einfluss 400 sqq.). Similarity of imaginal elements 
would be disturbing and ridiculous: lips like cherries, eyes like almonds, 
a skin like milk and strawberries, her beauty hangs upon the cheek of 
night as a rich jewel in an Ethiop's ear — in every case it is the feeling 
that is the decisive element, not cognitive similarities. The feeling is 
intensified and expands in fresh, strongly emotive words, in order to 
re-inforce and enhance the original emotion. The poetic figure exists 
to intensify emotion, not to supply the reader's imagination with more 
material (Miiller-Freienfels refers to Ribot, Essay sur I'imagination 
creatrice, and to Betz, Vorstellung und Einstellung, Arch. f. d. g. 
Psych. 17, 20). 

These problems would seem to require much further research. 

11.54. The Metaphorical Experience. Stahlin (Met. 25 sqq.) has 
investigated the nature of the metaphorical experience, and I shall 
summarize some of his statements, making use of my own terminology 
and the results of the preceding argumentation. 

The metaphor aims at providing relief and expression for the speak- 
er's feelings, and at impressing the hearer in a definite way. The 
metaphor carries along with it from its own sphere (the sphere of the 
primary referent) elements of its primary meaning. The strength and 
obtrusiveness of these elements depend on the circumstances, and on 
the degree to which they are applicable to the actual referent. The 
essential point is that we experience the actual meaning of the meta- 
phorical expression, and of the context in which it is placed, and si- 
multaneously also something of the primary meaning of the phrase 
or word. The fusion of the latter elements with the actual meaning 
and the actual context constitutes the metaphor. There is always a 
certain tension between the actual context and the primary meaning. 
The latter comes from an alien sphere — according to Stahlin, cf. 11.42 
— and in its whole import could not be applied to the actual referent: 
a man is not identical with a lion. But characteristics are evoked, 
relations and correlates occur to us, emotional tones make themselves 
felt, and we realize a certain Sphdrenhewusstsein: all this we are con- 
strained to apply to the actual referent, and to link up with the actual 
context, which thus becomes the bearer of these foreign characteristics,. 

308 GUSTAF STERN 11-54 

relations, and emotional tones. The blending of the two spheres, the 
application to the actual referent of characteristics not properly be- 
longing to it, of emotional tones of an unusual kind, of relations of a 
startling and original nature — this constitutes the essence of a metaphor. 
"Es findet ein Austausch der Merkmale, eine Vereinigung der beiderseit- 
igen Sphare, eine Verschmelzung von Bild und Sache statt" (1. c. 28). 

The metaphor expresses shades of thought and feeling which cotdd 
not otherwise be formulated in speech, or not so concisely and precisely 
formulated. The value of a metaphor lies in the adding of new attri- 
butes to a referent, its placing in a net of new complex relations, through 
which it is brought into a new light, receives peculiar emotional 
values and is comprehended more vividly and completely than before 
(1. c. 47 sqq.). With regard to images, Stahlin thinks that they serve 
to attract the attention to the primary meaning and thus prevent the 
swift comprehension of the actual meaning, as 'they may stress certain 
elements of the primary meaning which must be abstracted from in order 
to allow the actual meaning to establish itself (1. c. 33 sqq.; cf. 4. 141). 

The metaphor is not an abbreviated simile, for it does not compare; 
in calling a camel the ship of the desert we do not compare a 
camel to a ship — the common element is not sought (cf. 11.53, and 
Elster II 119). Metaphor is thus a sort of verbal shorthand, express- 
ing thoughts which otherwise could only be expressed through a long 
and cumbersome phrase. It gives the emotion directly, instead of talk- 
ing of it; it does not describe, but makes us experience. "Die Meta- 
pher is eine schone Kiihnheit, leicht verstandlich dem, der Geist hat" 
(Vischer, Das Symbol). 

A good metaphor is one which without effort moulds the apprehen- 
sion of the referent in a certain way, and thereby enriches it. (i) It 
must be effortless. (2) It must enrich the apprehension of the referent. 
(3) The fusion of referent and illustration {Bild) presupposes a certain 
affinity of their respective spheres. The nature of this affinity has not 
been investigated. (4) The metaphor must be such that the various 
attributes of the referent can coalesce into a unit in our mind.^) 

1) "Like all similitudes, Metaphors may (i) aid the Understanding, (2) intensify 
or work up a Feeling or Emotion, (3) give an agreeable Surprise. For the second 
and the third effects, and also as a distinct aim, they are required to be in full 
Harmony with the subject" (Bain I 159. See also K. O. Erdmann 172 with a 
quot. from Schiller, 210, 214 sqq.). 


The metaphor may fade, so that it loses its metaphorical force, no 
longer attracting associations from its original sphere. It no longer 
creates a tension, but becomes the ordinary appellation, the plain state- 
ment, for a referent: the head of a hand of robbers (see 14.52, adequa- 
tion). On the other hand the metaphorical elements of a phrase, its 
primary meaning, may occupy our minds to the exclusion of the actual 
referent, or at least so as to make it insufficiently comprehended. We 
do not realize that the expression is intended to be understood figura- 
tively. The metaphor misses fire, and is understood literally. Be- 
tween these two extremes we have the metaphor proper (Stahlin, Met. 


11.55. The Relation between the Referents. Most writers seem to 
take it for granted that the relation between the primary and the ac- 
tual meaning in a metaphor is always one of similarity, but when the 
term is defined as above, we get a comparatively large number of in- 
stances founded on other relations: a touch of shame upon her cheek, is 
a case in point (Bain I 167). The notion of shame is not similar to 
the notion of blushing, but correlated to it. 

It might be objected that the basis of a metaphor like this is a simi- 
lar feeling-tone (cf. the quotation from Miiller-Freienfels 11.53), and 
it is often not easy to decide how an individual instance should be 
judged. However, I need only refer to instances given in the lists 
below (11. 7). Since there are unintentional changes of two kinds, 
founded either upon similarity or upon other relations (regular trans- 
fers, and permutations, respectively), it seems reasonable to assume 
that just as we have an intentional type corresponding to regular trans- 
fers, so there must be one corresponding to permutations; a shift of 
this kind must also be capable of being effected intentionally in con- 
nection with emotional factors. 

It seems probable that metaphors founded on similarity are more 
frequent than the other kind. In a speaking community the opinion 
on what is similar and what is not, must be comparatively uniform. 
A similarity is therefore easily perceived, even if the metaphorical ex- 
pression is taken from an alien sphere, or involves only unimportant 
characteristics of the actual referent. But the apprehension of what 
is correlated in other ways, among disparate notions, must vary con- 
siderably from speaker to speaker, and from instance to instance, de- 
pending on personal experience and attitude, and on the momentary 

310 GUSTAF STERN 11-55 

circumstances. We cannot, except among intimates who know one 
another's ways of thinking, go very far in the chain of correlates without 
finding so great a variety that no two persons would be likely to hit 
upon the same, and the risk for misunderstandings would grow in the 
same degree. Among intimate friends the case is different: we know 
that in every little group of speakers there are allusions that are in- 
comprehensible to outsiders "not in the know". It should not be 
forgotten that metaphors founded on similarity may be so far-fetched 
that they are incomprehensible without a previous knowledge of their 
referent. Keats' lines in the Ode on a Grecian Urn, already quoted, 
are an instance. Many riddles are founded on elusive similarities. 

11.56. Hyperbole. I have defined hyperbole as a figure of speech 
in which the enhancement (see 11.15 and 11.42) takes the form of 
an exaggeration, an enlargement, a multiplication, or an intensification 
of the referents denoted; in other words, in the use of stronger words 
than the referent actually merits. Hyperbole is often combined with 
metaphor, so that a word is both hyperbolical and metaphorical (/ 
have oceans of time). "Hyperbole is an effect gained by magnifying 
things be3^ond their natural bounds. . . When an object pleases greatly, 
in consequence of certain qualities, we are willing to purchase an addi- 
tion to the pleasure, by raising or intensifying the verbal description 
of these qualities. This is Hyperbole, or exaggeration used for effect 
in style. The essential conditions are — (i) that the pleasure be marked 
and decided, (2) that the departure from truth does not shock our 
sense of the truthful, (3) that the language used be able to sustain 
the emotional interest" (Bain I 225). 

Bain further states that "our pleasure in overstating whatever con- 
cerns our feeling is so great that our ordinary language contains hun- 
dreds of hyperboles, which we employ without being tired of the repe- 
tition: not knowing the right hand from the left, splitting hairs, cheese- 
parings, over head and ears, driving a coach and six through an Act of 
Parliament, a sea of faces, speaking volumes" (1. c. 228, cf. Elster II 194, 
Nyrop IV 108 sqq., Carnoy 357 sqq.). Most of Bain's instances are 
hyperbolic phrases rather than hyperbolic words, and the sense-change 
is therefore mostly occasional only. 

But there are a great number of instances where the enhancement 
lies in the use of some single word, expressing a great intensity or a 
high degree, an enlargement, etc., which may correspond to the mo- 


jnentary state of jnind or purposes of the speaker, but which have no 
objective foundation in the actual characteristics of the referents and 
in their emotional value to the speaker. 

Among these words we may distinguish two types. In the first type, 
the enhancement follows, so to speak, the same line of thought, only 
pushing higher up along it: the primary meaning of the hyperbolic 
word is not disparate from the actual referent. Instances are: an endless 
day, for a long day, a thousand apologies, for many apologies. The same 
effect may be attained by adding a qualifier like very, much, highly, 
indeed, or other words of the type described in 14.531, in which the 
cognitive import has more or less completely faded, so that the word 
only serves to re-inforce the meaning of whatever other words it is 
used to qualify. 

I do not think that this kind of hyperbole involves an immediate 
sense-change of the type discussed in the present chapter. The word 
must be assumed to render the speaker's intentions, or momentary 
state of mind; it is true that the hearer may take it at a discount, but 
that does not constitute a permanent transfer, even in the case of re- 
petition. On the other hand, if the use becomes habitual, emotional 
adequation, as described in 14.532, may set in, the hyperbolic word 
losing more or less of its force; but that is another type of change. 

The second type of hyperbolic words is constituted by those in which 
the primary and actual meaning are disparate; that is to say, by words 
that are both hyperbolic and metaphorical. To this type belongs the 
large number of intensifiers that have not yet been completely ade- 
quated, but retain more or less of their primary meaning, and which, 
for that reason, give more force and vividness to speech, and are there- 
fore resorted to when greater expressiveness or effectiveness are desired: 
lovely, delightful, splendid, glorious, superb, grand, wonderful, gorgeous, 
heavenly, sublime, magnificent, perfect, divine, tremendous, entrancing, 
stupendous, enchanting, or in colloquial speech words of more or less 
"slangy" character: bang-up, rattling, tip-top, and so on. If much used, 
such terms succumb to adequation, and pass over into the former type 
(cf. Greenough & Kittredge 312 sqq.; French instances in Carnoy 357 
sqq.. Bally II 210 sqq.). In this type, there is a sense-change connected 
with the use of the word as applied to a disparate referent, that 
is to say, with the metaphorical use. 

As stated above, I do not distinguish hyperbolical from non-hyper- 

312 GUSTAF STERN 11-56 

bolical metaphors. Thus, in the following lists of words, many are 
hyperbolical, as for instance, a forest of feathers (ii. 6ii a), beak for 
nose (1. c), corn-stalk for a lean person (ii. 6ii b), Venus for a beautiful 
woman (ii. 6ii c). Paradise for a pleasant place, and Tophet or Ge- 
henna for an unplesant one (ii. 612 g), mountains of infamy (11. 612 a), 
iceberg for an unemotional person (11. 612 b), ancient damnation (11. 
612 c), pig, bear, and others (11. 612 d), carrion for a base or worthless 
person (11. 613 c), and many others. See especially 11. 613 for meta- 
phors based on similarity of emotive effect. 

11.57. Litotes, (i) The litotes, sometimes called meiosis, is de- 
scribed by Bain (I 232) as being the negation of the opposite to that 
which is meant, or, as Fowler expresses it (MEU), that particular 
kind of rhetorical understatement in which for the positive notion 
required is substituted its opposite with a negative: a citizen of no mean 
city, er freut sich nicht wenig, je ne vous blame pas. The last instance 
is from Nyrop (IV 268) who defines litotes as a rhetorical figure "par 
laquelle on attenue I'expression de sa pensee de maniere a laisser 
entendre le plus en disant le moins", and which mostly takes the form 
of negation of the opposite of that which is really meant. "In i Cor. 
xi. 7.22, / praise you not has the effect of an emphatic I blame; not 
a few means a great number, not bad, eh? after an anecdote, means 
excellent" (MEU, s. v.). 

When Fowler calls such expressions understatements, I think he is 
not quite right. Not bad, taken literally, leaves a large latitude, from 
indifferent to excellent, and may mean either, depending on the intona- 
tion used and the circumstances.^) As this figure does not give rise 
to any sense-change, it need not detain us further. 

(2) The term litotes is also used for an understatement "not to de- 
ceive but to enhance the impression on the hearer . . . taking many 

^) Fowler MEU 383 speaks of this figure as a "faded or jaded elegance . . . 
jaded by general over-use; faded by the blight of worn-out humour, with its not 
a hundred miles from, not unconnected with, and other once fresh young phrases. 
But the very popularity of the idiom in English is proof enough that there is 
something in it congenial to the English temperament, and it is pleasant to believe 
that it owes its success with us to a stubborn national dislike of putting things too 
strongly. It is clear too that there are contexts to which e. g. not inconsiderable 
is more suitable than considerable ; by using it we seem to anticipate and put 
aside, instead of not foreseeing or ignoring, the possible suggestion that so-and-so 
is inconsiderable". 


forms and contrasted with hyperbole. Very common in colloquial and 
slang English. The emphatic rather (Did you ever hear Caruso? Rather), 
the retort I'll see you further first {i. e., in hell)^), and the strangely 
inverted hyperbole didn't half swear (swore horribly), are familiar in- 
stances" (Fowler, MEU 6io). German instances are ein trinkharer Wein, 
ein leidlicher Verdienst. (Wellander, Studien I 192). 

Here too, the literal meaning may be intended, and the listener will 
have to depend on context, and signals, to interpret the speech 
correctly. The typical characteristic of this type is the use of a modi- 
fying attribute, denoting literally a lower degree, but intended to convey 
the notion of a rather high degree of whatever is denoted by the quali- 
fied word. The adjectives in -ish belong here: that will be a longish 
job, which may be stronger than that will be a long job. Similarly, / 
kind of got that habit myself, may imply that the speaker had the habit 
very badly (Bogholm 123). Cf. Swedish ganska Idngt, with emphatic 
stress on the qualifier. — This type leads to no permanent sense-change. 

(3) A third type of litotes consists in substituting a word denoting 
an inferior size, quality, etc. of whatever is being mentioned, as the 
little village, for London, the herring-pond for the Atlantic, my diggings, 
for, possibly, my quite comfortable apartments, and my old cart, for 
my brand-new Rolls-Royce. The reason for this is sometimes a humor- 
ous intention, sometimes a desire to avoid even the semblance of 
bragging of one's own belongings (cf. the quotation from MBU 
above). It might be called an inversion of the hyperbole. In this 
type the litotes is metaphorical, and there is consequently a sense- 
change involved, which may become permanent. The instances do 
not seem to be very numerous: herring-pond, and diggings, or den, are, 
I think, instances in point. 

The principle underlying the effect of litotes, as that of irony, has 
been formulated by Bain as follows: "It is a first principle of the 
htunan mind, that we are affected only by change of impression. Among 
the many consequences of this law is the efficacy of contrast in verbal 
composition ... It is a prevailing habit of language to express only 
one term of these couples [i. e., of correlated and contrasting notions) 
and to leave the other to be implied or understood" (Bain I 196). 
Cf. the quotation from Shand (11. 41) on the effect of surprise. The 
contrast involved in a litotes is that between the moderate emotional 

^) This is in my opinion a typical euphemism. 

314 GUSTAF STERN 11-57 

intensity of the actual expression, taking it literally, and the enhanced 
real meaning of that expression. 

For the speaker, we have to assunie incentives siniilar to those in- 
volved in metaphors: the desire for relief to one's feelings, through an 
emotionally coloured expression, and the striving for effect. 

With regard to the process of comprehension, the hearer has to educe, 
with the help of context, what the speaker really means. As in meta- 
phorical speech, the hearer knows enough about the topic and about 
the speaker's attitude to the topic to be able to conclude that when 
somebody calls London the little village, or the Atlantic Ocean, the 
herring-pond, or his fine car, my old cart — that in these cases the 
speaker is greatly understating his real opinion. The contrast between 
what is actually said, and reality, is generally humorously intended, 
and is consequently aiming at a stylistic effect. 

As stated, it is apparently only the third type of litotes that leads 
to a real sense-change. The non-metaphorical litotes might involve a 
secondary sense-change through adequation (14.54). 

11.58. The Classification of Metaphors. According to my plan, I 
confine myself in the following lists to metaphorical words, leaving out 
of account metaphorical phrases. 

With regard to the metaphorical words, I have already stated in 
1 1. 3 1 that I propose to distinguish metaphors founded on similarity 
from those founded on other relations. Secondly, I separate the various 
parts of speech. The nouns are, as usual, best represented. 

Among metaphors based on similarity we may, for the nouns, make 
distinctions according to the nature of the common element. The 
similarity may refer to (i) the appearance, (2) the qualities, functions, 
activities, or (3) the perceptual or emotive effects, of the referents. 
With a larger and more diversified material, it might have been possible 
to make further distinctions along these lines. 

Among metaphors founded on other relations I give only a few 
representative groups, not aiming at completeness. i) 

^) Stahlin (Met. 72 sqq.) proposes a very elaborate system for classifying 
metaphors, according to the sphere of experience to which the primary referent 
belongs. I am somewhat doubtful about the usefulness of this principle for seman- 
tic purposes. 


11.59. A Note on Marty's Theory of Metaphor. 

Since the late A. Marty's theory of metaphor has lately been taken up by 
Funke and Wellander, it is necessary to say some words in criticism of it. It is 
■closely involved in Marty's theory of figurliche innere Sprachform, a concept to 
which he attributes a great importance. Reviewers seem to regard it with some 
scepticism, tempered with respect for the formidable volume of Marty's writings 
on the subject, but as far as I know it has never been thoroughly criticized. 
I shall only offer some remarks, in so far as the matter is relevant to my own 
discussions. In summarizing Marty's views I follow Funke's book on Innere 
Sprachform, which makes use mostly of Marty's own formulations. 

Funke explains Marty's theory in the following way. "Lowe bedeutet das 
bekannte Tier; nun kann dieses Wort auch gebraucht werden, um einen Krieger 
zu bezeichnen, der sich als heldenhafter Kampfer hervortut. Wenn ich von 
jemandem in diesem Sinn als von einem Lowen spreche, so wird zunachst durcli 
den Namen die Vorstellung des kiihnen Raubtieres erweckt, durch diese Ver- 
mittlung aber erst die eigentliche Bedeutung, d. h. das, was eigentlich gemeint 
ist, namlich die Vorstellung eines Mannes von bestimmten Qualitaten. Es werden 
also in diesem Zusammenhang mit dem Aussprechen des Namens Lowe zwei Vor- 
stellungen wachgerufen: die zuerst ins Bewusstsein tretende dient aber nur als 
Band der Assoziation fiir die eigentlich zu erweckende Bedeutung" (Funke, 
Sprachform 26). The essence of the inner form is thus the occurrence of certain 
"Vorstellungen" which are called up by the word, but do not themselves con- 
stitute the meaning of the words; they only serve to evoke this meaning, accord- 
ing to the laws of association (Funke, Sprachform 27). In many cases the speak- 
er and the hearer do not consciously realize these elements, although they would 
be able to do so. The inner form has faded, owing to lack of attention to it. 
Habit ensures a direct and firm association between sound and meaning. But 
in other cases these elements of inner form are still actualized. The inner form 
"ist nicht das Bezeichnete, sondern selbst ein Zeichen, so gut wie der Laut" (1. c). 

This analysis should be compared to Marty's definition of meaning: "dasjenige 
psychische Phanomen, welches der sprachliche Ausdruck im Horer wachzurufen 
bestimmt ist" (Funke, 1. c. 20 — 21). It is clear that there is a contradiction 
between this definition and the assertion that inner form is not an element of 
meaning. The argument that inner form is not intended (bestimmt) to be evoked, 
but only the meaning is intended, falls to the ground, because if the inner form is not 
evoked at all, i. e., if no thought of the lion is evoked, then the expression is no 
longer a metaphor, but an ordinary unmetaphorical appellation of the referent. 

With regard to Marty's analysis of the comprehension of a metaphor, there 
are several objections. The first objection regards the tacit assumption that 
the metaphor {Lowe) is interpreted on the basis of the metaphorical word alone, 
as if it were isolated. The analysis of comprehension in ch. 6, and the further re- 
marks made above (11.52) will have shown that this is a serious mistake. Words 
are interpreted en fonciion du schema d'ensemble, and all relevant items of context 
are brought to bear in the process. Marty was, evidently, not in touch with the 
experimental psychology of his own time. 

3l6 GUSTAF STERN 11-59 

My second objection refers to the words, "so wird zunachst durch den Namen 
die Vorstellung des kiihnerO-) Raubtieres erweckt" — a phrase that begs the ques- 
tion. The point that requires explanation is just the fact that Lowe evokes the 
element of 'courage', among the many other elements of meaning that might 
have been evoked, and for all of which the name is equally a sufficient stimulus. 
Why does it not evoke instead the thought of four legs, or yellowish brown colour, 
or a tail, or nocturnal habits, or of a mane, or of great strength, or of canine 
teeth, or of roaring, or of the last lion I saw, which was in the Zoo? In Marty's 
analysis we find nothing on this essential point (cf. the explanation above, 11.52). 
If, as is mostly the case, the hearer knows what the topic is, the detour via das 
kuhne Raubtier is unnecessary. There are many instances in which it is absolutely 
necessary for the hearer to know to what referent the metaphor is applied, be- 
cause otherwise he would not understand it. I have already quoted in this 
respect Keats' lines in the Ode on a Grecian Urn. 

My third objection refers to the whole theory of mediation. We know from 
experimental investigations (cf. 6.22 and 6.36) that comprehension may be 
mediated by mental content of some kind. But the evidence does not show that 
metaphors and mediated comprehension are inseparable concomitants, or that there 
is any correlation at all between them, causal or otherwise. Such a correlation is 
the first thing that must be proved, if Marty's theory is to be taken seriously. 
(See also 14.15 on the theory of mechanization). 

A fourth serious objection may be made against Marty's method of trying 
to find the explanation of metaphors in the mental processes of the hearer. 
It is the speaker that formulates the metaphor, and its genesis can be explained 
only if we investigate the speaker's mind (11.51). The hearer has to adapt 
himself to the speaker's mood; there is nothing extraordinary in his methods 
for comprehending what is being said. For the speaker, the inner form cannot 
be a mediating element in the way Marty says. 

I conclude, then, that Marty's analysis of the comprehension of a metaphor 
does not agree with the findings of experimental psychology, as summarized in 
ch. 6. It is a mistake to assume that each word is interpreted as if it were isolated, 
neglecting the important part played by context. It is a mistake to suppose that 
mediated comprehension and metaphorical expressions are inseparably connected; 
at least no connection has as yet been proved. And it is a mistake to neglect the 
mental processes of the speaker, which are decisive for the psychic structure of 
the metaphor. The theory of inner form seems to me to be an unnecessary elabora- 
tion which is not corroborated by the results of psychological investigations. 

11.6. Metaphors based on Similarity. 

11.61. Nouns. 

11.611. Similarity of Appearance. This type is liable to produce 
numerous doubtful cases, merging into the non-metaphorical intentional 

^) My italics. 


transfers, and into regular transfers. The group is very large. The 
words are concrete in nieaning. 

(a) Object's Name for Object. Close at hand the building is an enter- 
taining salad of styles (1893). — / haue maintain d that Salamander 
(= fiery-red face) of yours with fire, any time this two and thirtie yeeres 
(1596). — An old turnip of a watch (1840). — The bay is now curling 
and writhing in white horses under a smoking south-wester (1849. Cf. F. 
moutons or chevaux blancs. Wundt II 606. — A Forrest of Feathers 
(1602). — The mounteins of bodyes were a-boute hem so grete that noon 
myght come to hem but launchinge (1450). 

Plants and flowers are often nanied on the basis of some observed 
resemblance; Weekley (Words 29) enumerates crowfoot, crane's bill, 
larkspur, monkshood, snapdragon, corresponding to G. Hahnenfuss, 
Storchschnabel, Rittersporn, Eisenhut, Lbwenmaul. Or are these in- 
tentional transfers? Parts of the hiunan body are not seldom denoted 
by words originally signifying material objects or more or less corres- 
ponding parts of an animal's body, the intention in both cases being 
humorous, often also depreciative: beak, bill, proboscis, or bow-sprit for 
nose, fin, flipper, paw, for hand, paunch, crop, craw, maw, gizzard, 
bread-basket for stomach or belly. 

{b). Object for Person. A tall and lean person may be called a bean- 
pole, beanstick, cornstalk, lamp-post, or rake. The secondary meanings 
of hulk, mite, clod, clown, are too well known to require explanation. 
A poker is a person with stiff rigid manner, a stick also 'a stupid or 
incompetent person'. Further instances, see A. Smith.^) The effect 
is generally depreciative. 

(c). Proper Names in Appellative Use. Proper names of persons 
are used as appellatives with regard to exterior appearance. We speak 
of a Venus, Helen, Adonis, Apollo, Goliath, Tom Thumb, Polyphemus, 
and so on. (Further instances in E. L. Fischer, Aronstein, Reinius, 
Ostberg, A. Peterson and Hvald Miiller). I may add that appellative 
personal names may be used of animals: "Adjutant, the nickname of 

^) From my point of view there is no purpose in distinguishing, as Smith does, 
the so-called synecdoche from the metaphors. When a person with fat or bloated 
cheeks is called chops, or a lean person skeleton, a corpulent or gluttonous person, 
guts, an informer or spy, a nose, etc., the circumstance that the words primarily 
denote parts of the human body makes no difference in principle. The last two 
instances belong to metaphors founded on similarity of activity or function, 11.612. 


the solemn Indian Stork, is clearly due to Mr. Atkins, and the secretary 
bird is so named because some of his head feathers suggest a quill be- 
hind an ear" (Weekley, Words 34). 

11.612. Similarity of Quality, Activity, or Function. With a larger 
collection of instances this group would perhaps split up into several, 
although there would always be numerous doubtful cases. It is some- 
times difficult to distinguish from the preceding group. 

{a) Object's name for Other Object. Concrete for Concrete. A mazev 
laherynth of small veines and arteries (1615) — During this autumnal 
season the city is a desert (1862). — Vsing alwaise soch discrete moder- 
ation, as the scholehouse should be counted a sanctuarie against feare (1568). 

Concrete for Abstract. This group comprises an enormous number 
of figurative abstract uses of concrete words, mostly in texts of a poet- 
ical or otherwise emotional character. The great frequency of the 
figure is due among other things to its being a convenient way of 
expressing what could not otherwise be so effectively put. A concrete 
mould, so to speak, is pressed on the abstract notion, thereby increasing 
its vividness and the force of its appeal to the listener. 

How now Ther sites? what lost in the Labyrinth of thy furie? (1606). — 
He would sleepe securely vpon the lap of God's protection (1617. Cf. for- 
tune's lap, the lap of luxury) — The savage creed that wears the scalp of 
Shelley at its belt (1870). 

A transfer of this kind is especially easy with regard to words denot- 
ing or implying some concrete quality which is, metaphorically, predic- 
ated of abstract entities. Such words are height, breadth, depth, colour, 
balance, support, root, stem, fruit, and many others. See the lists in 
Bain I 164. The favour of a king can remove mountains of infamy (1771). 

— A whole forest of verbal arguments. — Let no proofe be brought for it, 
and neuer so much against it, yet stickes the scarre of suspition still (1583) 

— The Prosecution would leave a Scar upon his good Name (17 10). — 
Waters of cursyng . . . muste be cast out of ^oure pytt with a scope of pen- 
aunce (1440). — The fende wil pluk at the balance To wey vs doun 
(1384). — Beir equal ballanis baith to riche and puir (1573). — The other 
characters are all sawdust and wires (igo8). — Our hereditary nobility 
have safety-valves in their rank (1818). — The barke ^at defendeth the 
tree from stormes and tempestes is hope. And the sap that gyueth lyfe to 
both, is charite (1526). — Now opyn yowur sachell with Laten wordis, 
Ande sey me ^is in clerycall manere (1450). — - That which we call Rail- 


levy, in This Sense is the very Sawce of Civil Entertainment (1692). — 
Fooles, that can not search the leakes of his defectes (1602) — He . . . suff- 
reth vs as for oure exercise With sharp e scourges of Aduersitee fful ofte 
to be bete in sondry wise (1386). — Pe lombe her lantyrne withouten drede 
(1360). — Yboundyn in the blakke barke of care (1374). — Her synd eac 
Pa cnihtas . . . mid dam ic becom to cristes scole (1000). 

Abstract for Concrete. "The abstract name isolates the point of im- 
portance, and so gives it emphasis. 'Youth' means the young, con- 
sidered as young" (Bain I 184). The following instances are from Bain: 
Let not Ambition mock their useful toil (Gray). — Fair star of the evening, 
splendour of the west. — Fold A rose-leaf round thy finger's taperness 

Abstract for Abstract. He had silenced the discords of passion in 
his own breast (1878). — Coveting no other companion but sorowe, nor no 
other harmonic but repentance (1588). These two instances might 
perhaps be classed under 11. 613. — Only those who have worked out 
their own photographic salvation can realize the difficulties to be over- 
come (1891.) — Fis psalme contens pe sacrament of all chosen men (1340). 

"Knowledge is light, passion is fire, depression of spirits gloom. . . 
So we speak of a ray of hope, a shade of doubt, a flight of fancy, a flash 
of wit, ebullitions of anger" (Bain I 163). 

[b) Object for Person. The name of an object is used for a person 
mostly in a depreciative sense. The quality denoted or implied is 
good or indifferent in the object, but is, or is considered to be, depre- 
ciative or insulting when ascribed to a human being, for whom another 
standard is required. The number of words used in this way is very 
large (see A. Smith, for additional instances; the following words are 
taken from his lists). 

My friend the noble captain — the illustrious general — the bladder 
(1840). — A tragedy queen, and a brimstone to boot (1824). — Hang off 
thou cat, thou bur (1590). — Dr Sanders the Popes firebrand in Ireland 
(1583). — We neither care for devout dunces nor for intellectual icebergs 
(1882). — They look'd Toward their chief and mouth-piece (1805). — 

And in the same way stick, sponge, sieve, block, satellite, sapling, 
blunderbuss , screw, wall-flower, and a host of other words. 

(c) Quality for Person. Shakespeare says "Bring in the admiration" 
(the admirable person), and Polyxenes calls Perdita, enchantment. Ju- 
liet calls her nurse ancient damnation. In modern English we speak of 


a failure^ a fraud, tenor, success, an awful warning, an inspiration ^ 
despair, hope, dependence, aversion, ruin, destruction, salvation, love, 
joy, delight, horror (Bain I 190, Greenough & Kittredge 257). Some of 
these may be permutations, cf. 13.223. 

I include here the use of titles such as His Majesty, His Grace, His 
Holiness, Your Excellency, if they are metaphors; they may be inten- 
tional transfers, non-figurative. But the following is clearly meta- 
phorical: "Up goes my grave impudence to the maid" (Bain I 184). 

{d) Animal for Person. These are often depreciative, more or less 
abusive appellations of human beings. The element of similarity is 
either a quality that is reprehensible or contemptible in itself, or else 
a quality that is neutral or favourable in an animal, but becomes re- 
prehensible in a human being. Wundt (II 605) reckons in this group 
(Metapherworter) "in erster Linie eine grosse Menge von Schimpf- 
wortern: so die Uebertragungen von Tiernamen auf den Menschen, 
wie Esel, Rind, Schwein, Affe, Gans, u. s. w. . . . Generell ist nur die 
Neigung, iiberhaupt Tiernamen als Schimpfworter zu gebrauchen, 
eine Neigung, die sich hauptsachlich aus der dienenden, ganz von der 
Willkiir des Menschen abhangigen Stellung erklart, die das Haustier 
in dem menschlichen I^eben einnimmt" (cf. Carnoy 325). 

Slopper found him a species of barnacle rather difficult to shake off 
(1868). — Such preetie Begles (= beagles) haue these Bishops . . . That 
hunt out Prebendes fatte for them (1570). — The French people of learn- 
ing . . . are not bears as most of ours are (1751). — Oh you beast. Oh 
faithlesse Coward, oh dishonest wretch (1603). — Ulysses looking sourly 
answered, You Bitch (1675). — 

And similary fox, pig, vixen, cat, snake, serpent, and many others. 
Lion is one of the few that is used in a favourable sense. (See Gree- 
nough & Kittredge 363). 

{e) National Name {used for Individual) in Appellative Use. She 
was of a wild, roving nature, inherited from father and mother, who were 
both Bohemians, by taste and circumstances (1848). — / never knew your 
grandmother was a Scotchwoman: Is she not a Tartar too? (1663). — 
Additional instances, see Reinius 169. Nyrop IV 381, quotes F. spar- 
tiate as denoting a person of rigid and firm character, which is in accord- 
ance with historical evidence, while beotian for a dull and stupid 
fellow is said to be due to an ill-natured quip on the part of a neigh- 
bouring province. Nyrop also quotes F. Suisse. There was formerly 


in Paris a Swiss guard distinguished by a peculiar uniform. A similar 
uniform was worn by the gatekeepers of palaces belonging to the no- 
bility, and these gatekeepers were often also Swiss. Hence Suisse is 
used to denote 'gatekeeper, concierge'. But this is explicable as an 
adequation or permutation. 

The present type is mostly depreciative, but Greek and Roman may 
be appreciative (cf. Carnoy 274 and Schreuder 96). 

(/) Proper Name in Appellative Use. As the instances show, these 
shifts are either appreciative or depreciative, and the qualities referred 
to may be exterior. The name is often preceded by the indefinite 
article. The following instances are too well-known to require docu- 
mentation: Croesus, an admirable Crichton, Xantippa, Job, Jezebel, 
Judas, Shylock, Lothario, Mrs. Malaprop, Delilah, Darby and Joan, 
John Bull, Lovelace, Hector, Amadis, Celadon, Don Juan, Solon, Solo- 
mon, Cicero, Benjamin, Joseph, Benedick. Jonah causes a storm. Pan- 
dora brings misfortune, Ganymede and Hebe wait at table, Figaro is 
a barber. Mentor and Egeria give good counsel, Jehu has his counter- 
parts in modern life, and so has Nimrod. 

In some cases the proper name has passed altogether into the class 
of appellatives: martinet, pander. The origin of hooligans is (according 
to Weekley, Words 12, and COD), a spirited Irish family, "whose pro- 
ceedings enlivened the drab monotony of life in Southwark towards 
the end of the 19th century". Magdalen has given Maudlin, which 
was applied to people who were weeping, tearful, lachrymose, in allu- 
sion to pictures in which the Magdalen was represented weeping (Week- 
ley, Words 61, and ISTED. See the lists in Ostberg, Reinius 20, Nyrop 
IV 100, 364, A. Peterson, and Ev. Miiller). 

I reckon in this group names of relations used as appellatives: father, 
nuncle, gaffer, gammer, daddy, granny, brother, sister, gossip, are used 
in an extended sense of other persons showing some qualities in com- 
mon with the original possessors of the name. At first the use was 
probably humorous or endearing (Reinius 186). 

Proper names are sometimes more or less fancifully used of persons 
without any discoverable likeness between primary and actual referents . 
It is comprehensible that Jack, owing to its great frequency, becomes 
equivalent to 'man'. (Ostberg 73 and 76). Tom, Dick, and Harry, is 
= 'everybody, the mob'. In other cases the appellation goes back to some 
real or fictitious personage: Black Bess, Peter Rugg, Peeping Tom, Paul 

Goteb. Hogsk. Arsskr. XXXVIII: i. 21 


Pry, Anthony Cufjin; an Irishman is Paddy, a Welshman, Taffy, a 
man from the Tyneside, Geordie, and so on, because these names are 
common in the respective districts. Persons of rustic manners are 
called by common rustic christian names, as Hodge, Hick, Hah; and 
women: Betty, Jenny, Mawks, Molly. Betty and Molly are also used 
contemptuously of a man who occupies himself with a woman's house- 
hold duties; Miss Nancy is a finikin, effeminate man (Ostberg 49, 86, 
et al.) Slang and cant make an extensive use of christian names in 
this way. 

(g) Place Names in Appellative Use. Our house is a sort of Bedlam, 
and nothing in order (1713). — Every province . . . would in turn appear 
a Paradise, and a Pandaemonium (1779). — Bridewell, from {St.) Bride's 
Well, in lyondon, is "a place of forced labour, a gaol". Babel, Mecca, 
Capua, Golconda, Tophet, and others, are well-known instances (see 
Efvergren) . 

11.613. Similarity of Perceptual or Emotive Effect. The common 
element in this group is the similarity of perceptual or emotive effect 
on the subject. We touch here on one of the most difficult problems 
connected with metaphors, and very little can be said with any 
pretensions to* definiteness. 

(a). Synaesthesia. Synaesthesia is an association that connects elements 
from different sensory spheres {Sinnesgebiete) , the point of similarity 
being constituted by their effect on the perceiving subject. (Cf. Sie- 
bold 5 sqq.). 

It is somewhat doubtful what the precise nature of the common ele- 
ment really is. Spearman (242) states that the awareness derived from 
the several senses must at first always be perfectly alike, but this fact 
has perhaps no bearing on the matter in hand. Wundt (II 553 — 554) 
regards synaesthesia as a case of "complication" between "die ein- 
fachen Empfindungen und die an sie gebundenen sinnlichen Gefiihle". 
and considers it due to the fact that disparate sensory qualities evoke 
similar feelings; a word denoting one kind of sensory impression can 
therefore be apprehended as an adequate name for an impression 
from another sensory sphere. 

According to Sterzinger, synaesthesia is explicable if we assume that 
the point of similarity is a configuration, and he regards synaesthesia 
as "das Auftreten von Vorstellungen aus anderen als den primar erreg- 
ten Sinnesgebieten durch Vermittelung oder auf dem Umwege iiber 


eine zwischensinnliche Gestalt" (Sterzinger, Arch. 37, 386 sqq.). These 
intrasensorial configurations Sterzinger considers as identical with the 
"unspezifische Momente oder Elemente" spoken of by Ruths (Kxp.- 
unt. iiber Musikphantome, 1898, 196 sqq.). Some configurations are 
related to what Biihler calls Regelbewusstsein. Cf. Siebold 37. 

Synaesthesia is especially common among adjectives (see 11.62), 
but there are numerous instances of nouns in E. von Siebold's lists (to 
which I refer; see also Elster I 387): the sound and light of sweetest songs 
(Swinburne), The stars were the notes of the singing, and the moon was the 
voice of the song (James Thomson). 

I have no instance of synaesthesia effecting a permanent sense-change 
of a noun. 

(b). Abusive Words as Endearments. An abusive word may be used 
in a sense that is the opposite of its primary meaning: heau'nly foolel 
(Sidney, 1586). In these cases, a preceding process of adequation, in- 
volving disappearance of the cognitive elements of meaning (see 
14.531), is apparently not to be assumed. The shift is effected, in- 
tentionally,, in one movement. In some cases, we are able to point 
to a specific quality that excites the emotions of the subject. Thus, 
a wretch is often an object of pity and commiseration, because he is 
persecuted or otherwise unfortunate. But other small, weak, defence- 
less, things, who are not 'wretches' in the primary sense, are also ob- 
jects of pity, and so a connection may be found to mediate a shift. 
Witch is used of 'a young woman or girl of bewitching aspect or man- 
ners'. For terms like lamb, kitten, mouse, the smallness, softness, play- 
fulness, or rather their impression on the speaker, are more or less 
prominent: trusty turtle trewest of al trewe (14 . .). Ej, My dere lady, 
My mouse my nobs (: darling) and cony (: rabbit) swete (1567). God 
blesse thee Mouse the Bridegroome sayd, and smakt her on the lips (1586) . 
Cf. Kdtzchen, Mauschen (Wundt II 575 sqq.), mon petit lapin. 

In other cases, no connection between the referents except the emo- 
tive element can be detected: pure terms of abuse may express friend- 
ship. The following is an instance, rather extreme but not uncom- 
mon: "First Youth: Hullo, congenital idiot! Second Youth: Hullo, 
you priceless old ass! The Damsel: I'd no idea you two knew each other 
so well! (Punch). Cf. German Luder, Aas, Schelm, and others, as 
terms of endearment. 

Wundt (II 576) assumes that an abusive term as such possesses asso- 

324 GUSTAF STERN 1 1 -613 

ciations which give it a tendency to shift from hate to love. In the 
first place we have to think of the intensity of the feeling as such, with- 
out regard to its quality or direction. "Wie uns schon in den Benen- 
nungen der Affekte durch die Sprache iiberall die Tatsache entgegen- 
tritt, dass fiir die unlustvollen Affekte iiberall reichere und intensivere 
Bezeichnungen zu Gebote stehen als fiir die lyustaffekte, so verbindet 
sich auch in der affektvollen Rede mit dem Gebrauch gesteigerter Aus- 
drucksformen der Uebergang auf die negative Gefiihlsseite, die um 
ihrer grosseren Intensitat willen im positiven Sinne verwendet wird". 
The instinct for giving to the feelings as strong an expression as possible 
makes us resort to these words. The designations for unpleasant emo- 
tions are not only more numerous, they are also of a greater intensity. 
When we wish to express a very strong pleasant emotion, an expression 
in reality belonging to painful emotions easily presents itself as the 
most adequate symbol, adequate to the intensity, although disregard- 
ing the quality, of the emotion. In this process speech only follows 
the course of the emotions themselves, in which, when we have attained 
the highest intensity of joyous surprise, the sudden transition to pain- 
fulness follows, evoked by the concomitant physical processes. The 
change of abuse into endearments belongs to the instances in which the 
striving for an adequate relief to the emotion calls for the name of the 
opposite emotion. The feeling for the contrast thus evoked, and the 
feeling of humorous irony towards paradoxical expressions, may also 
be contributory factors. Wundt states that furchtbares Gliick and 
schreckliche Freude also belong here, but I think that we have to make 
a distinction between words where a preceding process of adequation 
is likely to have pushed aside the cognitive elements of meaning, and 
instances where no such process has occurred, and where the meta- 
phorical shift is therefore possible only intentionally, and as a more 
forcible deviation from ordinary usage. 

Otherwise, Wundt is probably right in his explanations. The pivotal 
element in the change, the point of similarity, is evidently emotive, 
and the intensity of the emotion is the characteristic that evokes a 
word corresponding to it. Intense emotion is liable to inhibit other 
mental processes, and to attract all attention to itself; the disregard 
of the cognitive elements of the meaning is thus explicable, even with- 
out assuming a previous adequation. Context must also be considered, 
as usual. The feelings of the speaker for his or her interlocutor are 


such that any expression of emotion must be an endearment: the 
listener is in no doubt whatever on that point. The facial expression, 
tone of voice, and the whole behaviour of the speaker, often contri- 
bute to the effect, and provide additional context for guiding the inter- 

(c). Appreciative or Depreciative Uses. A third group, related to 
the preceding, is constituted by the use of objects' names for persons. 
Here, too, there seems to be no previous adequation with disappear- 
ance of the cognitive elements. Thus in the following instance: Lady, 
Pat clept art modir of mercy, Noble saphir (15th c); and similarly are 
used the words gold, gem, jewel, diamond, peach, daisy, cherry, honey, 
and others. As in the preceding group, it is sometimes a specific quality 
of the referent that mediates the shift, as in the use of lily for denoting 
persons or things of exceptional fairness or purity: The name of seinte 
Cecile ... it is to seye in englissh heuenes lilie, For pure chastnesse of 
virginitee (1386). 

Similarly, >vords denoting vile or disgusting objects are applied 
opprobriously to persons: It were better for a woman to be barren. Than 
to bring forth a vile wicked carren (: carrion. 1547). That dirt of a cap- 
tain (1658). See A. Smith for further instances. 

11.62. Adjectives. Most adjectives denoting qualities of concrete 
objects can be used metaphorically of abstract things. The instances 
are innumerable, and I shall only give a few, as illustrations. A larger 
collection of material would perhaps show how to distinguish smaller 
groups, which I cannot do now. 

I dradde after so noble spekers . . . to putte for^ my bareyn (: barren) 
speche (1387). — You may easily perceiue what successe they are like to 
haue, that deale with so leaden and sandie braines (1590). — And he to 
me wit thow maid ony fait. To the that wil be ful sowre and salt (1500) . 

— Well agyd now, but sappy strength he kepes of grener yeres (1558) . 

— / am not saplesse old or reumatick (1598). — Blowe hence these sapp- 
lesse jestes (1602). 

Synaesthesia is, as I stated above, very common in adjectives. I 
quote the following instances from Stern, Swift 272 sqq. Light is used 
of food 'that does not lie heavy on the stomach, easy of digestion'; of 
wine 'containing little alcohol'; of medicines and other objects that 
may be apprehended as having an activity of their own: 'mild, gentle'; 
of immaterial objects: 'easy to bear or endure'; of temptations: 'mild. 

326 GUSTAF STERN 1 1. 62 

gentle'. Sharp is used of scourges and other objects that cause a pain 
similar to that caused by a sharp instrument, 'painful'; of weather: 
'stinging, keen, intense, violent, vehement'; of a sound: 'penetrating, 
shrill'; of the rays of the sun: 'piercing, intense'; of words: 'cutting, 
severe, harsh, stern, peremptory'; of hunger, sorrow, etc.: 'stinging, 
intense, severe'; most of these senses are also found in smart. Cf. Lat. 
acutus. Qviick is used of the voice: 'loud, clear'. Similarly lijlic. We 
speak of warm and cold colours, and warm and cold friendship or reception. 
(Cf. Stocklein 73, Carnoy 288 sqq., Siebold 102 sqq., Noreen, Betr. 283). 

11.63. Verbs. As with regard to adjectives, I give only a few in- 
stances in illustration of the general practice of using concrete verbs 
in abstract meanings: 

They . . . gonnen wade in meny an uncouthe glad and depe mater e, As 
jreendes don, whan they hen met yfere (Chaucer 1374). — Some few mer- 
chants and tradesmen . . . may not grind them so as shall always keep them 
in poverty (1626). — / am obliged to 'grind' , that is, undergo a private 
examination with an authorized teacher or tutor (1835). — Neither stuff e 
the bodye, nor choke the conceit, which it lightly doeth, when it is to much 
crammed (1581). — The sodayn dreade of euery bodely payne woundeth 
vs to the hearte and striketh our deuocion starke dead (1534). — For it 
neyther savoureth the spirite of God, neither yet any modest and good na- 
ture, hilt . . . (1574). — Hate teres of gretyng, pat ^e synful sal scalden in 
pe dounfallyng (1340). — Nowe a wheale on suche noses . . . That so 
quicklie canne sente where hidden golde dothe lye (1553). — Her eyes were 
saluted with a tuft of trees (1586). 

11.7. Metaphors based on other Relations. 

This type corresponds to permutations, and it is often difficult to 
discriminate the two. The same shift may no doubt be sometimes 
effected in either way, intentionally as a metaphor, or unintentionally 
through a permutation, and we do not always know the history of 
each word well enough to place it definitely. 

Most of my instances are nouns. Adjectives, adverbs, and verbs, 
are so scantily represented that it does not seem worth while to class 
them separately. It is not impossible that, at least for adjectives and 
adverbs, this circumstance reflects a factual rarity of such metaphors 
in speech, but more research is necessary before a final verdict can be 


11.71. Article of Dress, Tool, Implement, etc. for Person. The earlier 
shifts of this kind give one the impression of being mostly permuta- 
tions (see 13.218), but among later instances there are certainly many 
metaphors, since they are clearly used for reasons of style (humorous 
intention, etc.). I quote the following (from A. Smith). 

Never reached by any other spiritual organizations except possibly the 
poke-bonnets at the corners of the streets (the Salvation Army women). 
■ — Mr. Blower, the eminent silk-gown (King's Counsel). — Sweet-meats 
being great favourites with the pig-tails (Chinamen). — Two good knives 
and forks (hearty eaters). — Bung is a nickname for the master's as- 
sistant who superintends the serving out of the grog. Pen, for writer, 
may be a permutation: What harme may hap by helpe of lying pennes 
{1563). A book wherein a second Pen had a good share (1605). But very 
often it is used with a stylistic purpose, and is then a metaphor. 

As metaphors we should probably regard the plural type of such 
appellations: guts, buttons, breadbags, chips (a ship's carpenter. Kip- 
ling), chops, gloss-eyes, marrow-bones, pudding-sleeves. The earliest 
instance seems to be velvet-guards, trimmings of velvet, and hence the 
wearer of such trimmings: Sweare me, Kate . . . A good mouth-filling 
Oath: and leaue in sooth. And such protest of Pepper Ginger-bread, To 
veluet-Guards, and Sunday-Citizens (Shakspeare 1596). The reason for 
the plural form was that each person possessed the thing in plural 
number (cf. Nyrop IV 195, Carnoy 260). 

11.72. Symbol for Thing Symbolized. When we say crown, sceptre, 
or throne, for 'royalty', the reason is stylistic, and consequently the 
shift is classed here. Other similar expressions are mitre, strawberry 
leaf, altar, pulpit, hearth, bench, bar, wool-sack, etc. (Cf. Nyrop IV 190 
sqq.) Ceres is used for bread, Bacchus for wine, and similarly Mars, 
Neptune, Pallas, Venus. Grey hairs symbolize age, Arabia, spices, etc. 
The power of the purse, the cottage, the palace are other well-worn instan- 
ces. "These symbols appeal to the senses, and help to make the subject 
more impressive" (Bain I 187). Some of the words adduced in the 
preceding paragraph may fall under the present heading. Lawn, or 
lawn-sleeves, is used not only for a bishop, but also for the office of 
bishop: You ask me if I ever knew Court chaplains thus the lawn pursue 
(1732). A man of great Note for the sake of Laun-sleeves is aturning his 
Coat (1710). Cf. 13.214, where corresponding permutations are listed. 

11.73. Material for Object. The name of a material may be used 

328 GUSTAF STERN 11-73 

metaphorically for objects manufactured from it: steel may designate 
a weapon of steel; "the marble speaks" (Bain I 193); laurel msLj stand 
for laurel wreath: Laurear of martirs, foundid on holynes (1430). Cf. 
permutations of the same type, 13. 211. 

11.74. Proper Name in Appellative Use: For Persons. The London 
Policeman is called a bobby, formerly also a peeler, both from the name 
of the founder of the corps, Sir Robert Peel. Tommy Atkins, the nick- 
name of the Army man, is said to be from the official blanks in which 
the fictitious name of Thomas Atkins was inserted to show the recruits 
where they ought to fill in their own names (Ostberg 77). 

Possibly some of the names in 11.612 f should rather find a place 
in the present section. 

For Animals. A jack-ass is a male, and a jenny a female of the species; 
similarly nanny-goat and billy-goat, tom-cat and tibby or tib-cat; jenny- 
cat, jack-rabbit, etc. Poll is the conventional name for a parrot, and 
hence designates any parrot (N^D). The proper name combined with 
the animal's own designation generally denotes the gender, and the 
use of it alone may possibly be a shortening, according to Nyrop IV 
361, who adduces Renart goupil > Renart, and compares Robin mouton, 
Jeannot lapin. Further instances, see Ostberg 89. 

For Objects. These shifts of proper names are sometimes, as far as 
can be gathered from the available evidence, "irrational", and should 
perhaps be classed in 11.78. I have provisionally placed them here. 

The ISTED enumerates a great number of mechanical contrivances 
of different kinds which are all called jacks, and suggests that the name 
was applied to things which "in some way take the place of a lad or 
man, or save human labour", but adds, "also more vaguely of other 
things with which one has to do". Judging by the instances quoted, 
there is no real proof of the assumption that jack was at first used 
for things which more directly take the place of a man. The application 
of the same name to many different animals and flowers appears equally 
fanciful. Probably there is some reason and connection at the start,, 
but what that is lies entirely outside our knowledge (cf. Ostberg 99). 

Jenny is also used for machines. The NED suggests that since Jenny 
is a feminine of Jack, it was used as a feminine prefix and as the name 
of machines. Again, the instances give no guidance for testing the 
assumption (cf. Reinius 143 sqq.). Davit, for machines, especially for 
the contrivance for lowering ship's boats, is from David, and the reasoiL 


for its use in this way is uncertain. Jemmy, pet-form of James, for 
"a crowbar used by burglars, generally made in sections screwing 
together", is another unexplained shift. The same tool has earlier been 
called hess or hetty. In German it is Dietrich, Peterchen, or Klaus, and 
the contracted forms of the first name, dyrk, dirk, have passed into 
Swedish and Danish with the same meaning (Weekley, Words 42; 
see further Ostberg 98 sqq. French instances, Nyrop, Etudes 33). 

The history of grog is more roundabout. It was first a nickname 
for Admiral Vernon, owing to his wearing a coat of grogram. When he 
decreed, in 1740, that rum was to be served out mixed with water, the 
sailors gave his name to the mixture (NED). Gladstone is used for a 
cheap claret, because Gladstone reduced the duty on French wines 
(Ostberg 94). Drinking-vessels are not seldom provided with humorous 
or endearing names. See Ostberg 100. 

11.75. Dates for Events. The Forty-five, is not only the year 1745, 
but also the Jacobite rebellion of that year. Le Quatorze Juillet is 
employed as equivalent to the revolution of 1830. 

11.76. Habitual Expressions for Persons. Habitual expressions 
may be used for the persons who employ them. Thus, the name 
of jingoes, for "the new tribe of music-hall patriots who sing the jingo 
song" (1878), is from the asseveration by jingo in a popular song of the 
period, when war with Russia was threatening. I^ater on the word has 
been extended to ultra-patriots of other times and other countries 
(Weekley, Words 13, and NED). 

Nyrop (IV 210) states that in Paris slang an Englishman is called 
un goddam or un goddem; the form un godon is found already in the 
15th century in Normandy; similarly in Spanish un godon. All these 
forms are from the oath goddam. A parlez-vous, for a Frenchman, is 
quoted by the NED from 1815. (Further instances, see Nyrop 1. c, 
Carnoy 263, and Reinius, Onomatop. Bezeichn.). 

11.77. Place-names in Various Uses. 1 select only the following 
illustrative instances from Efvergren, to whom I refer for further 

Philistia is used for "the class or community of Philistines, i. e., 
unenlightened or commonplace people, or the locality they inhabit". 
Buncombe, or bunkum 'political claptrap', from an episode in Congress, 
when the member for Buncombe, speaking unnecessarily, declared 
that the people of his constituency expected it, and that he was bound 

330 GUSTAF STERN 11-77 

to make a speech for Buncombe. Grub-street, the name of a street "much 
inhabited by writers of small histories, dictionaries, and temporary 
poems", hence used metaphorically for the "tribe of mean and needy 
authors or literary hacks". 

11.78. "Irrational" Metaphors. Metaphors are sometimes founded 
on an entirely arbitrary and fanciful connection, a witticism, or pun. 
I have already pointed out that some of the instances in 11.74 may 
belong here. A typical instance is tandem 'at length', applied to the 
harnessing of two horses one behind the other, instead of side by side. 

Jespersen (Punning and All. Phrases) lists a number of such meta- 
phors, mainly real or fictitious place-names. I'm for Bedfordshire 'I 
want to go to bed', and with similar meaning, to go to Ruggins' ; to 
send a person to Birching-lane 'to whip'; he got it by the way of Cheapside 
'he got it cheap'. To travel by Mr. Foot's horse, to go by the Marylebone 
(marrowbone) stage, to ride shanks' mare, all mean 'to walk'. (Cf. also 
Efvergren 109 sqq., and Carnoy 378). 

11.8. Euphemism. 

11.81. Definition, and Formal Types. I have defined euphemism as 
a tendency to tone down or veil dangerous, indecent, or otherwise 
unpleasant things, and also as the linguistic result of this tendency 
(Cf. 10.184, 1142).^) 

The various linguistic forms which a speaker may employ owing to 
a euphemistic tendency, may be described as follows (see especiall}^ 
Nyrop IV 262 sqq.). 

(i) Clipping or omission of the main word. A French instance is 
jarni = je renie, with Dieu left out. Nyrop also speaks of aposiopesis 
of an offensive word, which description covers many of the instances 
given by Wellander, Studien II 9 sqq. English instances are Zounds for 
God's wounds, where only the -s is left of the "dangerous" word; and 
d — ox d . . . , for damn or devil, sometimes "pronounced" dashed or 

^) See further Jespersen, Growth a. Str. 242, Carnoy 337, Hatzfeld, L,eitfaden 
107, Stocklein 40. Nyrop (IV 257) states that euphemism "designe maintenant 
I'ensemble des moyens linguistiques par lesquels on deguise une idee desagreable, 
odieuse, ou triste. L'euphemisme est done un adoucissement, grace auquel on 
evite le mot propre dans les cas ou son emploi pourrait choquer pour une raison 
quelconque." He remarks that "reuphemisme est dicte tantot par la decence, 
la politesse, la prudence, tantot par une certaine crainte superstitieuse; on y 
recourt par egard pour soi-meme ou pour les autres" (1. c. 260). 


deed (Jespersen, Gr. a. Str. 243). Nyrop lists this type separately as 
abreviation. In My! and Oh my! the offensive word is left out alto- 
gether. See also 10.184. 

(2) Deformation or distortion of the offensive word is closely related 
to the preceding type. Under this heading come the well-known and 
well-worn blooming for bloody, dash, darn, dum, GosJi, Golly, Gorry , 
Cox my passion, by cock, cock's bones, odd's bodikins. Gad, and many 
others, which need not be translated (cf. Swaen, Engl. Stud. 24). 
Nyrop's substitution (1. c. 264) in which a word with different meaning, 
but similar in sound, is substituted, is related to what I have called 
"irrational" metaphors (11.78). II a de la menthe dans son jardin = 
il mente. It may be classed here. 

(3) A foreign word is substituted, which, being less definite in meaning, 
is in the same degree less offensive. Beckman (131) explains this by 
the circumstance that the foreign word is more or less a blank, which 
obediently conforms to the use we make of it, while the native word 
has numerous undesirable associations with the offensive thing, and 
means exactly what it means (cf . Cederschiold, Kvinnosprak 99). Another 
circumstance to be noted is that the foreign word is sometimes of a 
more general meaning, as described in the next group. Instances are 
cliemise for shift or smock, lingerie for underclothes, effluvium for stench, 
perspiration for sweat, intoxication for drunkenness, maniac for madman, 
dipsomaniac for sot, etc. (Greenough & K. 306, Nyrop 1. c. 264). The 
use of an obsolete word of the same language is preferred on the same 

(4) A term of vague or general import is used instead of the more 
precise word. Flannels or linen for underclothes, smallclothes for 
breecJies, and similarly in various senses misconduct, misguided, misde- 
meanour, offence, fault, slip, lapse, transgression, excesses, etc. These 
last words may also be classed in nr. 6 (Greenough & K. 305). 

Nyrop (IV 266) points out that in this way a word of general import 
may receive two or more specific euphemistic meanings. In lyC medecin 
malgre lui, Sganarelle says to a woman that her husband deserves 
qu'elle lui mette quelque chose sur la tete; il a eu quelquechose entre eux 
means that there has been a dispute; il a pris quelquechose, that he has 
been severely trounced. We are reminded of the corresponding use of 
German etcetera, adduced by Wellander (Studien II 13): ich hiess ihn 
.einen et cetera; sie fuhre ein liederliches Leben, und gebe ein offentliche 

332 GUSTAF STERN 1 1. 8 1 

& ccetera ah. The word may mean 'podex, crepitus ventris, devil, cacare*^ 
etc. This might be called a pronominal use. 

(5) Flattering words, or words denoting the opposite of that which 
is meant, are sometimes used (Nyrop 1. c. 267: noms flatteurs and 
antiphrases). Blessed is used for 'cursed': every blessed evening. The 
results of taboo often belong to this type, which touches upon irony. 

(6) lyitotes is sometimes euphemistic, and it is only natural that 
the two figures should be closely related, especially the form of litotes 
that consists in negating the opposite of that which is meant (11.57). 
Untidy or unclean is used for dirty, untruthful for lying, intemperate 
for drunken, unwise for foolish, unsafe for dangerous, impolite or 
uncivil for rude, etc. (Greenough & K. 307). I am not sure that all 
these words are really euphemistic; at least the NHD gives no indication 
of such a shade of meaning for them. Nyrop gives French instances 
that appear more convincing: "une vieille femme n'est plus feune, une 
piece ennuyeuse nest guere amusante" , etc. In Hnglish one speaks of 
a lady of a certain age, the discipline leaves much to be desired, and so on. 
We do not like to say, that is a lie; we say instead, / am afraid that you 
have been misinformed, there must he a misunderstanding. A long- 
winded orator tries to keep his audience in good temper by assuring 
them that he is only going to say a few words more; the man who will 
he back in half a second employs the same method. 

Slang and cant are full of these euphemisms: even if a thief is speak- 
ing only to his comrades, he will prefer not to say that he is going to 
steal: he is going to do a job. Corriger la fortune is a polite French cir- 
cumlocution for cheating at cards (cf. Cederschiold, Kvinnosprak 
78 sqq.). 

(7) We have, finally, the clearly metaphorical euphemism. To this 
type belong, among many others, to go to a better world, to join the 
majority, to go west (Jaeger 289), and other expressions for dying. In 
Australia a transported convict used to be called an old hand (Gree- 
nough & K. 305). Sent up means put in prison; to draw the long bow, 
is the same as giving a free rein to one's imagination, which latter phrase 
perhaps also belongs here. A fairy tale is a lie; to be d sport has often a 
distinctly unfavourable meaning. A rest camp is a cemetery (war slang; 
on this see Jaeger). 

11.82. Causes of Euphemism. One kind of euphemism originates 
in ancient taboo. The name was regarded as an integrating part of a 


being, human or otherwise, and the use of the name was believed to 
attract the being itself, which might lead to fatal consequences in the 
case of evil and powerful spirits. Cf. the German proverb Wenn man 
den Wolf nennt, kommt er gerennt, and the Swedish Ndr man talar om 
trollen sd dansa de i farstun, 'when you speak of the witches, they are 
dancing at the door'. This belief led to the avoidance of the real names 
of gods, demons, and other dangerous beings, and the use of numerous 
circumlocutions and metaphors. The bear is 'the brown one', the wolf 
in Swedish varg 'the outlaw'. The most famous example is the Hebrew 
name of God, which was not to be used by any one. (Cf . Giintert, Die 
Sprache d. Gotter, Beckman 142, Oertel 304). 

The knowledge of the name was believed to confer to some extent a 
power over the thing or person named. Solomon's power over the 
spirits was founded on his knowledge of their real names. Although this 
superstition has still a firm hold on many people — perhaps not least 
in England — euphemisms of this kind are mainly traditional, or due 
to an unwillingness to sin against what is considered good taste, by 
using words forbidden in polite society; that is to say, euphemisms of 
taboo merge into the second type. 

The second type of euphemism may be termed the euphemism of 
delicacy or good breeding, and consists in avoiding that which might 
fall beneath one's own dignity, or below the level of style that we adopt, 
or which might wound the feelings of the hearer, or sin against the pre- 
vailing standards of decency or good behaviour. To this type belongs 
the euphemism of decency, which seeks new and indirect appellations 
for things, or actions, that are not considered mentionable in polite 
society: lavatory, unmentionables, and many of the instances already 
quoted above, innumerable names for the state of intoxication, for 
parts of the body, for articles of clothing, and so on (Weekley, Words 
98, Jespersen, Growth a. Str. 245). Some of these euphemisms should 
perhaps be considered as the result of various styles being used. If 
two medical men in a discussion use the word podex, that is not a 
euphemism, but a technical term the use of which belongs to the style 
they employ. But if some one else, in a discussion otherwise carried 
on in untechnical language, says podex — that is a euphemism, because 
he does it to avoid the use of an English term. 

A very large group of euphemisms is formed by expressions intended 
to put something in a way that shall not wound the feelings of the hearer. 

334 GUSTAF STERN 1 1. 8 2 

In speaking to some one who has suffered a loss by death, we do not Hke 
to use the word die, but prefer to say pass away, go to fiis last rest, and 
so on; cf. departed, removed, fallen asleep, if anytliing sliould fiappen, 
(Bain I 183). Such expressions are resorted to by a speaker for his own 
sake (/ cannot hear to speafi of it), so as not to harrow his own feelings 

The numerous euphemisms for 'kill' may be of either kind: to mafie 
away with, to put away, to finish, to settle, do for, remove, lie must disap- 
pear, etc. (Greenough & K. 303). They may be the result of a desire 
to present what the speaker is doing in a favourable light. Cf. mafie a 
job and corriger la fortune, quoted above. Many people will say lend 
me a fiver, although both parties know that give me a fiver would be 
more adequate. 

In order to spare a person's dignity, we will say that he is stout, plain, 
or deranged, rather than fat, ugly, or mad. This is especially the case 
when we are speaking to a person; cf. the already quoted you have 
been misinformed, and similar expressions. 

Generally speaking, the causes of euphemism are to be sought in 
the striving of the speaker to adapt his words to the effect which he 
wishes to make on his interlocutor, an effect which might be said to 
consist in the avoidance of certain unpleasant associations. We have 
now to discuss the manner in which this effect is attained. 

11.83. The Psychic Process. What is the characteristic feature of 
the process of speech when we say pass away instead of die? 

Taking first the case of the listener, it may be said, generally speaking, 
that the word indicates the referent, assisted, as usual, by context, 
and further, that the word, being metaphorical, carries with it associ- 
ations from its primary meaning, giving a certain colour to the actual 
referent. The effect of this is to tone down too glaring colours, to 
present offensive things in a more delicate manner. The explanation 
of this ability on the part of euphemistic expression is, I think, to be 
sought in the factors that have been described above in analysing 
metaphors proper (11. 5). 

The habitual, direct appellation, for instance die, is used of the re- 
ferent in all contexts and situations. It is associated with the referent 
in all its aspects, even the most sordid or painful. A person who has 
just suffered a loss through death is naturally very sensitive on that 
point; the painful aspects of dying will be prone to arise in his mind 


on the least provocation. But if instead he hears death spoken of as a 
passing away, a going to one's rest, a less disagreeable side of the refer- 
ent is presented to him: a peaceful departure, a leave-taking for eternal 
rest. Such a term is not used in connection with the idea of painful 
death-struggles, putrefaction, worms, and other disagreeable concomi- 
tants of death. These associations are kept in the background, and in 
consequence the mentioning of the topic will be less liable to wound 
the feelings even of a sensitive person (cf. Carnoy 338). 

The effect is thus double: the avoidance of unpleasant associations, 
and the suggestion of less unpleasant, if not positively pleasant, ones. 
The speaker has often at his disposal more than one euphemism, and 
may select the one that best fits the occasion. 

Euphemism is often a matter of tone and manner. A person 
speaking of indecent things may do so in such a cool, dispassionate 
manner that the effect is one of euphemism even if the plain names 
of the objects are being used. On the other hand, a formally euphemistic 
expression, spoken in a suggestive, insinuating, unpleasant manner, 
may appear the height of cynicism. Cf. the well-known line in Faust: 
1st iiber vierzehn Jahr dock alt — which formal euphemism, but actual 
cynicism, revolts even Mephistopheles: Du sprichst ja wie Hans Lieder- 

In the case of pass away and similar euphemisms, the required tone 
of voice will not be cool, dispassionate, but consoling, compassionate. 

Euphemisms are evidently employed not only for the sake of the 
interlocutors, but also by the speaker for his own sake (cf. Nyrop, 
quoted above). An indecent, vulgar expression would not fit into the 
style of speaking employed by an educated person of some refinement, 
accustomed to keep his expressions at a reasonably high level of style. 
It would wound his own sense of dignity, as much as the susceptibilities 
of his hearers. In many cases, however, the reason for avoiding a cer- 
tain expression is nothing better than an exaggerated respect for Mrs. 
Grundy. (Cf. Jespersen, Growth a. Str. 245, and Fowler, MEU, s. v. 
Genteelisms) .^) 

^) Wellander (Studien I 190) sets euphemism in opposition to hyperbole, which 
latter involves a strong emotion. "Der Euphemismus entspringt der ruhigen 
Ueberlegung; der Redende will jeden Affekt vermeiden und weicht dem Ausdruck 
jedes starkeren Gefiihls aus: daher der allgemeine, abstrakte, bloss andeutende 
Charakter, der meistenteils dem euphemistischen Ausdruck eigen ist". A glance 

336 GUSTAF STERN 11.83 

The analysis in 11. 51 applies in many respects to euphemisms, and 
need not be repeated in detail. 

11.84. The Sense-change. With regard to the sense-change involved 
in euphemisms, we have at present only to discuss the transfer that 
may follow immediately, not the adequation that may ensue on habi- 
tual use. 

The last type described in 11. 81 (nr. 7), being metaphorical, involves 
an intentional transfer, as described for metaphors. As for the other 
types, nrs. i and 2 may be left aside with a reference to ch. 10. Nr. 
3 scarcely involves an immediate sense-change, if we keep to the strict 
letter of my definitions. We have to assume that chemise, when first 
used for smock or shift, really expressed a different way of apprehending 
the referent — otherwise it would not have been euphemistic. With 
repeated use, adequation may set in, but on that point I refer to ch. 
14.55. The same arguments apply, I think, to nr. 4, and perhaps also 
to nrs. 5 and 6, but the matter requires further investigation. 

11.9. Irony. 

According to Bain (I 213) "irony consists in stating the contrary of 
what is meant, there being something in the tone or the manner to 
show the speaker's real drift . . . Job's address to his friends is ironical: 
'No doubt but ye are the people, and wisdom shall die with you'. So, 
also, are the words of Elijah to the prophets of Baal: 'Cry aloud, for 

at the instances above is enough to show that euphemisms are the product of 
emotion; if that term is extended to cover also stylistic purposes, all euphemisms 
are emotive. Even if it is taken in a narrower sense, Wellander's contention is 
clearly not correct: the speaker who tries to hold his audience by assuring them 
that he will only say a few words more, may throw as much feeling into his words 
as any hyperbolical orator. The friend who comes to console one that has suffered 
a loss, and speaks compassionately of the dear one that has passed away, does 
not speak unemotionally; but on the contrary, with deep emotion. The point is not 
to avoid all emotion; the point is to avoid emotions of a certain kind, and this 
is done, as the instances show, either by substituting another shade of emotion, or by 
unemotional speech. The three adjectives allgemein, abstrakt, andeutend, by 
which Wellander seeks to characterize euphemisms, cannot be accepted without 
considerable reservations. Andeutend may suit all euphemisms, but allgemein 
fits only some of them; I refer to my instances. And why abstrakt} Surely 
wohlgendhrt is as concrete as dick, and die Speise nicht behalten cannot very well 
be less concrete than the verb for which it stands (the instances from Wellander, 
1. c. 191). 


he is a god; either he is talking or he is pursuing, or he is on a journey; 
or peradventure he sleepeth, and must be awaked' . . . Bentham, in 
his attacks upon the English Law, constantly describes it as our 
'matchless constitution' . . . The Mark Anthony oration is full of ironical 
effects. Such is the re-iteration of 'honourable men' applied to the 
murderers of Caesar". 

Irony is a typical case of indicated purport (see 4.265), in which the 
primary meaning of the word may be said to furnish one fundament, 
the relation of contrast is deduced by the hearer from context, and the 
second fundament is the actual meaning, at which the hearer arrives. 
The actual meaning is more or less the opposite of the primary meaning, 
and irony is thus the figure of contrast par excellence. 

If a speaker exclaims, yow are a fine fellow! with ironical intention, 
we have to assume that the context, perhaps especially the external 
context, the previous behaviour of the "fellow" that calls forth the 
exclamation, plainly indicates to the hearer that the speaker really 
means just the opposite. In oral speech, the tone of voice often assists 
comprehension; it may express indignation, or contempt, or any other 
emotion that is incompatible with the literal meaning of the phrase. 
But too much importance should not be attached to this point. Irony 
is sometimes all the more effective if spoken in a cool, dispassionate 
voice, and in written language intonation is often not indicated. It seems 
that the general situation is the most important factor for guiding the 
hearer to a correct comprehension: the irony is understood en fonction 
dn schema d' ensemble. Bain instances the sustained grave irony of 
Swift's Gulliver's Travels and The Tale of a Tub. 

The reason for the superior effectiveness of the ironical expression 
is the implied or expressed contempt or derision which must intensely 
annoy the person against whom they are directed, in addition to which, 
as Bain points out, it embarrasses an opponent by giving no opening 
for a reply. The speaker feigns agreement with his interlocutors, but by 
slightly exaggerating their claims he makes them seem absurd, and a 
legitimate object for scorn. He thus attains a stronger effect than by 
stating in so many words his real opinion: No doubt but ye are the people, 
and wisdom shall die with yon . . . The irony paves the way for the 
subsequent serious argumentation: But I have understanding as well 
as you. Sometimes the irony does not involve an exaggeration, but 
leads the hearers to conclude for themselves that the other party's 

Goteb. Hogsk. Arsskr. XXXVIII: i. 22 

338 GUSTAF STERN 1 1. 9 

pretensions are absurd. A conclusion which we make ourselves from 
the facts before us is generally more convincing than one which is forced 
upon us. Mark Anthony's speech is a good instance. The radical devia- 
tion from the habitual mode of expression contributes to the effect of 
irony, since a feeling of surprise draws the attention to, and intensifies, 
emotions associated with or immediately following it (see 11. 41). 

Turning now to the question of sense-change, we may take as a 
typical instance the exclamation you are a fine fellow! intended to 
mean just the reverse. Are we to assume an immediate sense-change 
for the adjective fine, just as we assume a sense-change for a word that 
is used in a metaphorical expression? 

According to my definition, the meaning of a word is the speaker's 
apprehension of the referent denoted by it, in so far as this apprehension 
is expressed by the word. Applying this definition, we have to say 
that the speaker's opinion of the referent is that it is 'not fine'. Further, 
it seems indisputable that the speaker denotes the quality in question 
by the word fine. But is this meaning, this notion, really expressed by 
/mg? The question is difficult, but I am inclined to answer it in the 
negative. It seems to me that the meaning is expressed indirectly, 
through an eduction of correlates. I admit, however, that the matter 
is by no means clear. 

On the other hand, if the ironical expression becomes habitual, 
adequation may set in, and the depreciative meaning becomes the real 
meaning of the word. See 14.56. 

Bogholm states that there are three classes of words that are especially 
liable to be used ironically, (i) Laudatory terms, as beautiful, fine, nice, 
precious, pretty: A nice job I have! Things have come to a pretty pass! 
Most precious nonsense! A fat lot of good it's done 21s! You're a beauty, 

(2) Old-fashioned and pompous words, which throw a comical light 
over the whole expression. He reproaches me with treachery, because 
forsooth I had not sent him a challenge! The fickle moon! quotha! I 
wish my friends were half as constant! I passed on and left him orating. 

(3) Intensifiers used ironically instead of down-toners. A lot you 
know about that! A similar instance is the use of much, as signifying 
'not much'. Falstaff speaks of men as their fathers' shadows, adding: 
hut much of their fathers' substance! Cf. the modern expressions, much 
you care! much he knows! and the curious not half! 


In addition to tone and facial expression, the ironical intention may 
be expressed by a changed word-order: Much you care ahoitt my feelings! 
(Bogholm, Bet. 116). 

I have to leave the matter with these remarks. Much additional 
research is necessary before we get a thorough grasp of the nature, 
causes, and effect of the figures. 


12.1. Theoretical Discussion. 

12.11. Analysis. We now come to what are sometimes called the 
"regular" types of sense-change, that is to say, changes that occur 
unintentionally in the course of ordinary discourse: transfer, permutation, 
and adequation. When necessary, in order to distinguish it from inten- 
tional transfer (11.3), the present class may be termed regular transfer. 

Regular transfer is the unintentional use of a word to denote another 
referent than the usual one, owing to some similarity between the two 
referents. The similarity in question is, at least for the momentary 
purpose, essential. 

Schematically, the transfer may be described as the use of a word 
habitually denoting the referent abed, to denote instead the referent 
cdef, because the elements cd for the moment are the only ones relevant 
to the context, and therefore predominate in the speaker's attention, 
to the exclusion of the elements ab and ef. In other words, cd are cen- 
tral elements of meaning (see 4.16), on which the meaning is speciaHzed 
(see 4.24). 

The psychological explanation of this type is similar to that of 
metaphors based on similarity (cf. 11. 6). Each referent has more 
than one characteristic; in so far as these are separately apprehended, 
there is in the meaning of the name a corresponding plurality of ele- 
ments. Not only the meaning complex, but each element of it, is associ- 
ated to the word. The thought of any separate characteristic of the 
referent, if sufficiently salient in attention, will be a sufficient stimulus 
for recalling the word. 

The momentarily irrelevant characteristics of the two referents, 
a b and e f, and the corresponding elements of the meanings, have been 
placed in brackets in the diagram on the next page. The identical 
characteristics cannot be assumed to be absolutely identical in the 
two referents, since in the one case they are apprehended as charac- 


teristics of P, in the other as characteristics of Q. I have therefore 
marked them c d and Cy d^, respectively, and the corresponding elements 
of meaning have been similarly distinguished. 

Word: X ^Y 

Meaning: (a^) yby^ 7i^i (eC) 

Reierent: {ah) c d c^ d^ {ef) 

-=P = Q 

If a speaker is talking of the referent P, and specializing on the char- 
acteristics c d, which for the moment are central in attention, while 
the elements a b are inhibited, the thought of the former elements, yd, 
may evoke in his mind, theoretically, either the word X, which is 
associated with yd, or the word Y, which is associated with the 
elements y^ 6^ (cf. the diagram and analysis in 11.51)^) 

It is the normal thing for a word to occur in a specialized meaning. 
An object, as for instance a brick, has various characteristics. As ex- 
plained in 4.24, a speaker using the word may be thinking primarily 
of the form and size of bricks, or of their consistency and resistance 
to pressure, or of their colour and exterior appearance, or of their 
function as elements in a building, or he may be thinking of them simply 
as undifferentiated entities. If then, a speaker perceives an object of 
brick-like shape, it will come natural to him to call that object a 
brick, because he is accustomed to call an object of that shape a brick 
without for the moment pa^dng any attention to the fact that real 

^) Stout (Psych. I 285 sq.) states that "it may happen that under conditions 
which would otherwise give rise to conflict, no appreciable conflict actually ensues, 
simply because either the old or the new combination is relatively so powerful 
as to overbear the tendency opposed to it without a struggle. Suppose the com- 
ponents of the one combination are abc and of the other abx. C may be favoured 
from the outset, so that it simply displaces x without any feeling of discrepancy 
arising, and without any attention to the difference. This process . . . may take 
place between a new percept and the predisposition left by previous percepts; 
when this happens, it may also extend to the corresponding memory-image, 
inasmuch as the modified perception gives rise to a correspondingly modified 

342 GUSTAF STERX 1 2. 1 1 

bricks have other characteristics too, which are not to be found in 
the actual referent; these characteristics are not attended to for the 
moment, and so cannot prevent the transfer. 

The transfer is thus intimately bound up with one of the fundamental 
facts of meaning: that we practically always mean, or refer to, a referent 
in a specific aspect, in which only certain elements of the semantic 
range are of interest, and function as central elements of meaning. The 
possession, by a referent, of numerous characteristics will present 
correspondingly numerous starting-points for transfers, as well as for 
other sense-changes. 

This explanation involves a restriction of the term transfer to denote 
shifts that are based on similarity. It has often been assumed that a 
word could be transferred to denote a referent standing to its primary 
referent in practically any relation. In my opinion, the material at 
present available shows that this is improbable. If we mean by transfer 
a shift of meaning that may occur unintentionally and without the 
mediation of a peculiar verbal context, the restriction is necessary. 
When the relation between the primary and the actual referent is not 
one of similarity, the shift is not so simple, and can occur only 
through the mediation of a peculiar verbal context, as described in 
the next chapter, or in the form of an adequation (ch. 14). For a 
transfer a specific point of view is required, but not a specific verbal 

The transfer is the simplest type of sense-change, which agrees with 
the fact that it is extensively used by children to supply the gaps in 
their vocabulary, aus Sprachnot. Since children often have only a 
vague idea of the real range of a word, their transfers are sometimes 
"wild", as compared with those of adult persons. 

The transfer is performed entirely by the speaker. The hearer's 
part is only to understand the shifted word, which will present no 
difficulties, since the shift is, by definition, so simple and self-evident 
that it passes unperceived. 

It is evident that there can be no equivocation (cf. 13.12) in a 
transfer. A hrick is either a brick of clay, or it is any other brick- 
like object to which the name has been applied, but the word could 
not possibly have an alternative meaning, as a word can in ^ per- 

A successful transfer, one which is repeated and adopted by the 


Speaking community, is followed by adequation (see 14.6), so that 
the word becomes capable of being applied to the actual referent even 
when the elements in which it differs from the primary referent — the 
elements ab in the diagram — are no longer inhibited, and even when 
they are predominant. i) 

12.12. Conditions and Causes. The conditions of transfer will be 
evident from the remarks in the preceding paragraph. A transfer can 
take place when a referent is apprehended in a particular aspect that 
makes only some of its characteristics stand out in attention, as central 
elements, and when, in addition, these characteristics are also to be 
found in another referent, the name of which can, consequently, be 
used for the first referent. As far as I can see, no particular verbal 
context is necessary, but only a specific aspect or point of view that 
makes some characteristics salient, while the others are inhibited. 
Possibly further investigation will be able to formulate conditions 
which we cannot at present perceive. 

As a favourable factor we have probably to consider the great 
frequency of all the common types of transfer. Every speaker is famil- 
iar with the method of transferring a word to another referent which 
has some characteristic in common with the primary referent. It is 
possible that some transfers "go in a series". Thus, when names for 
parts of the body are used for parts of a ship, perhaps a beginning was 
made with some obvious identity, as for instance side, and the series 
was continued with other names (cf. Falk 83). 

With regard to the ultimate causes of transfers, they should be sought 
in the functions of speech. By definition, it is the speaker's striving to 
make speech fulfil its symbolic and commtmicative functions that is 
active. Further, the process is unintentional. The new word offers 
itself spontaneously as a suitable expression for what the speaker 
wants to say, and is as spontaneously accepted, the speaker attending 
only to the content of his speech, leaving the form of it to the more 
or less automatic action of lower centres. 

1) The distinction I have made, between shifts based on similarity — transfers 
— and those based on other relations — permutations and adequations — corres- 
ponds more or less to distinctions made by some earlier writers, for instance by 
Wundt, between sense-changes with constant and with changing predominant 
element of meaning, and by Falk between changes based on associations of simi- 
larity, and those based on associations of contiguity. 

344 GUSTAF STERN 12.12 

The question why a new word is evoked and accepted, is a very- 
difficult problem. One reason may be Sprachnot, shortage of words. 
We should probably interpret the term as indicating not only a lack 
of words in the absolute sense, but rather as referring to the momentary 
supply or command of the vocabulary. It may happen that an other- 
wise perfectly familiar word does not arise at need, and it may then 
easily be replaced by another name for the referent. We have all ex- 
perienced the unpleasant sensation of not remembering a word, even 
if we direct our attention to the task of finding it; the same thing 
must happen also when we are not attending to the finding of words, 
and leave to the automatic processes the task of finding a substitute. 
And we have all heard speakers, without noticing the discrepancy; 
using words that do not at all fit into the context. If quite discrepant 
terms may displace the correct name, and pass off unperceived, it 
must be still easier for a new term that does fit the context to be unin- 
tentionally employed, and accepted without demur, both by speaker 
and hearer. 

It is possible that a word for some reason, recent use or frequent use, 
is more easily evoked than the ordinary term for a referent. Many 
persons have, in their daily work, a stock of words which are in constant 
use because they denote referents that one has constantly to do with; 
it is natural to assume that such words are favoured in comparison 
with others and have a lower liminal value, arising in mind on a slighter 
stimulus. Perhaps a word may be favoured because it has pleasant 
associations for the speaker. The possibilities are no doubt many, 
although we cannot at present enumerate them in detail.^) 

Whatever the reason for the individual speaker's choice of terms, 
it should not be forgotten that no sense-change results until what is 
at first an individual trait has become an affair of the speaking commu- 
nity or at least of some considerable group of speakers. We should 
therefore take into consideration only such motives as may affect a 
number of persons in a similar way. 

^) I do not think it is correct to ascribe the choice of one word in preference to 
another to the exclusive action of emotive factors, as Sperber does. Habit is 
not a feeling or the effect of feeling, but a tendency to repeat actions to which 
the organism has adapted itself more and more in the course of its own activities. 
Also, the daily work is for most persons an entirely unemotional affair; it is the 
change, the recreation, that is emotionally coloured (cf. 7.43 and Stern, Litteris 

in 59). 




There is nothing in the diagram in 12.11 to show why the transfer 
proceeds in one direction in preference to the other: in other words, 
why the name of the referent Q is used for P, and not vice versa. In 
many cases at least, the process apparently does work in one direction 
only, a circumstance that has not yet been investigated. Generally 
speaking, I think we may say that the name of a referent that lies closer 
to us, that is more familiar, or more concrete and palpable, or in which 
the quality that serves as common element is more salient, is employed 
to denote the referent that lies farther away, is more or less strange, 
or less concrete and palpable. This agrees with the assumption that 
the lower liminal value of some words may cause them to be used in 
a transferred sense. 

12.13. Delimitation and Definition. The question concerning the 
correct delimitation of transfers has already been touched upon above, 
with regard to nominations based on similarity. The diagram below 
will show how the classes IV — VII in my system border on each other. 




j^ Transfer 

•-H Figures of 

rt Speech 



•^ Intentional 
^ Transfer 

"^ Figures of 



I have defined the term intentional as signifying that the speaker's 
attention is directed, more or less consciously, to the form of his 
utterance, not only to its content (11. 13). Consequently regular trans- 
fers are characterized by the circumstance that the speaker is directing 
his attention only to the content of his utterance, leaving its form to the 
more or less automatic action of lower centres. In such circumstances, 
a transfer can take place only if the common element is essential to 
the referent. This gives us a theoretically definite criterion for the 
distinction between intentional and regular transfer. 

Further, regular transfers are often closely related to figures of 
speech founded on similarity. In order to distinguish them, I class 

346 GUSTAF STERN 1 2. 1 3 

as figures of speech all instances in which the speaker is seeking for a 
more adequate filling of the expressive and purposive functions of 
speech, including aesthetic (stylistic) purposes (cf. 11.42). 

I have consequently restricted the term regular transfers to instances 
where the motive power is the striving to fulfil as adequately as possible 
the symbolic and communicative functions of speech, and where there- 
fore no emotive factors are involved. I have no doubt that some scholars 
will want to move many of my metaphors over to the present group, 
while others would prefer to include the whole class of regular transfers 
among the metaphors. The question is partly one of expediency: we 
have to choose a distinction w^hich has at least some chance of applica- 
tion to facts; a distinction that is theoretically clear is sometimes 
inapplicable to the linguistic material. 

The regular transfers are distinguished from permutations and 
adequations by being based on similarity (cf. 13.14 and 14.14). 

There remains, finally, the distinction between transfers and fluc- 
tuations, perhaps the most difficult of them all (cf. 7.13). Fluctuations 
are variations of meaning which do not lead to "another" meaning, but 
stop within the range of the word. As a possible criterion, I suggest 
the fact that fluctuations keep within the same category, while a trans- 
fer does not do so. 

We should probably regard as fluctuations the variations in meaning 
of the word land in phrases like land and sea, land and water, land and 
people, or of the word state in state and society, state and people, state 
and church (Wundt II 537). On the other hand, if we employ the 
word saddle to denote a mountain ridge shaped like a saddle, the two 
referents do not belong to the same category. There are, as always, a 
large number of intermediate instances, difficult to classify. 

On the basis of these arguments, I define regular transfers as unin- 
tentional sense-changes based on similarity. 

12.14. Classification. As stated in 11. 31, I intend to follow the 
same principles in classifying transfers as I did for figures of speech. 
Consequently I begin by distinguishing the parts of speech, and within 
each of these groups I further separate the instances according to the 
nature of the common element; the identity on which the transfer 
rests may be an identity of appearance, of function, situation, etc. 
As usual, the material is scanty for adjectives and verbs. 

The transfers are exceedingly frequent in all languages, but since they 


are of a very simple type, they do not vary much. It is therefore not 
necessary to give so many instances; a few typical cases will suffice to 
give an idea of the nature of this kind of sense-change. 

12.2. Nouns. 

12.21. Identity of Appearance. We have, first of all, nouns denoting 
the simple geometrical and stereometrical shapes, with typical forms 
that strike the eye and furnish the predominant element of meaning, 
on which a transfer may rest. Ball is transferred to 'a globular or 
rounded mass of any substance', and to 'objects or parts with rounded 
outline'; similarly cone, cube, pyramid, tube, etc., and everyday objects 
of conspicuous shape, as stick, hoard, lane, vein, ladder, barrel, bowl, 
belly, bag, brick, leaf, ring, tongue; a bush is any bushlike bunch or tuft, 
a bear is 'a rough mat for wiping boots on, a block covered with shaggy 
matting, and used for scrubbing the decks of vessels' (or is this inten- 
tional?). With regard to bridge, how, saddle, and many others, not 
only the appearance, but also the function may serve to mediate a 
transfer. Thus, saddle may be used of 'some thing resembling a saddle 
in shape: a depression in a hill or line of hills; a long elevation of land 
with sloping sides, a ridge, esp. one connecting two hills; also, a similar 
formation of ice or snow', or it may signify 'a block on the top of a pier 
to carry the suspension cables; a 'seat' or support on which a gun is 
placed for bouching; a bracket to support the (telegraph-) wire on the 
top of a pole or ridge; and various other technical appliances'. 

A wafer is a 'light, thin, crisp cake, baked between wafer-irons'; or 
'a thin disk of unleavened bread used at the Eucharist', and the name 
is transferred to 'a small disk of flour mixed with gum and non-poisonous 
colouring-matter, or of gelatine or the like similarly coloured, which 
when moistened is used for sealing letters, attaching papers, or receiving 
the impression of a seal'; also to 'a thin leaf or paste, used to form a 
cachet for the administration of powder'. 

12.22. Identical Function. It is often impossible to decide 
whether identical appearance or identical function, or both, have been 
active in determining the transfer. A bonnet is 'a protecting covering 
or defence' in various technical uses; a bed is the 'last base or surface 
on which anything rests'; a bolster is used of 'various things in the 
nature of a pad, various parts of mechanism which form a solid support 
or base, on which other parts rest or exert pressure'; a bark is 'an outer 

348 GUSTAF STERN 12.22 

covering or husk'; a leg 'something more or less resembling a leg or 
performing its relative function as a support for a body'. In this case 
perhaps also the situation in a totality, under the 'body', may have 
contributed to the change. Bridle is used of various things resembling 
a horse's bridle in form or use. Mouth is 'the mouth of a river or bay' 
(cf. Norw. jjordkjcBJt, elvarkjcBJt, French houche). Wagon is used of 
many different kinds of vehicle. 

12.23. Identity of Relative Situation within a larger whole. When 
we speak of the legs, arms, feet, body, head, neck, etc., of an object, 
probably both appearance and relative position^) may be involved in 
the common element; with regard to jront, hack, top, bottom, sides, 
waist, and other similar terms, probably only the relative position. 
When we speak of the root of a hair, function and relative position have 
no doubt both contributed to the shift. The latter factor is perhaps 
alone in bough 'anything analogous to a limb of a a tree, in being a 
lateral extension or subdivision of a main trunk'; and in brow 'project- 
ing edge of a cliff or hill, standing over a precipice or steep'. 

12.3. Adjectives. 

12.31. Identical Appearance, Form, or Structure. Sharp may be 
used of the nose, referring only to the appearance, and losing the impli- 
cation of cutting or penetrating. Quick originally meant 'living'; it 
is used, perhaps owing to foreign influence, of coals, 'burning', and 
of water, 'running, flowing'. Fast 'immovable, inflexible, stable', is 
used of a covenant or law: 'firm, stable, reliable, not to be broken'; 
of friendship or fidelity: 'firm, steadfast'; of sleep: 'deep, sound, not 
easily broken'. (These instances are from Stern, Swift 272 sqq.). 

The adjectives denoting, simple shapes, as round, square, flat, narrow, 
low, and others, have a correspondingly simple meaning, offering few 
possibilities for specialization. The various uses are scarcely more than 
fluctuations within the range of the word. The application of such 
adjectives to abstract referents, as narrow mind, low cunning, flat notes, 
may be transfers or nominations, and some of them are international, 
so that foreign influence may have contributed to the present meanings 
in English. A detailed study of their history would be welcome. Similar 
observations seem applicable to the adjectives denoting colours. 

^) Relative position in a series is a characteristic of an object, just as much 
as form and colour are (Head, Aphasia I 527). 


With regard to these circumstances it is not unlikely that transfers 
are comparatively more rare in adjectives than in nouns: as pointed 
out above it is the occurrence of numerous characteristics in a referent 
that offers opportunities for sense-changes. 

12.32. Identity of Function, Ability, or Behaviour. Sharp is 
used of the sight or eyes, 'acute, keen', and of persons, their intellect 
or understanding, 'acute, keen, discerning'. The quality of being 
penetrating is here transferred to the mental sphere. Light is 
similarly transferred and is used of persons, their mood, mind, heart, 
or countenance, 'merry, cheerful'. The idea of being lightly burdened 
is here shifted to the mental sphere, and apprehended as signifying 
lightly burdened in respect of sorrows or cares, and therefore merry 
and cheerful. 

12.33. Relational Shifts. It is very common for an adjective to be 
attributed to a noun with which it is not logically compatible. We 
speak of the highest bidder, although, strictly, the bidder is not high, 
only his bid is. It is especially common with nomina agentis derived 
from verbs, the action of which is qualified by the attribute: an early 
riser, late corners, first offenders, he is a hard student, a long resident, 
a slow walker, we were both fair runners, a probable winner of the Derby, 
a close observer, a wide traveller, a habitual liar. Nouns which denote 
a person having a certain quality receive an attribute pertaining to 
the quality, not to the person: a perfect stranger, a total stranger, you 
are a positive fool, her particular friend. Adjectives of size are similarly 
used: a great admirer of Tennyson, he was an enormous eater. A noun 
of this kind may have two attributes, one referring to the person, the 
other to the action: an ordinary secret sinner, a young and rapid writer 
(cf. Jespersen, Grammar II. 12.12 sqq. with further instances). 

The illogical use of adjectives has been explained by Feldkeller in 
the following manner. A strictly logical syntax would require a measure 
of thought and attention which during the rapid course of speech is not 
always available, since we mostly concentrate our attention on the 
content of what we are saying, and have little time to spare for the 
form, which is left to the care of lower centres. These may guarantee 
the correct use of the basic meanings — they do not always, as shown 
above — which as a rule are sufficiently firmly associated to the words 
for mistakes to be comparatively rare if special reasons do not intervene. 
But a correct syntax is not so easily learnt. The speaker has often to 

350 GUSTAF STERN 12. 33 

formulate afresh, and cannot content himself with reproducing his 
stock of schemes. Mistakes therefore easily creep in, which the listener 
in his turn perhaps does not notice, since he is too much occupied with 
the content of the speech to pay attention to the syntax. The words 
give him the main points of what is being said; everything else he 
combines for himself, educing relations on the basis of the fundaments 
communicated — the basic meanings. Speech has preferred this 
method because it leads to satisfactory results with less effort. 

I only wish to add that we cannot class as mistakes all variations 
from strict logical correctness. Speech has always had a certain free- 
dom in this respect, and it is speech habit, not logic, that should be 
taken as the criterion of correctness. I admit that the limits of correct- 
ness are sometimes obviously exceeded, and that they are difficult to 
define with precision. When a German says ein niohlierter Herr, eine 
yeitende Artilleriekaserne, and the like, protests are not unjustified 
(Further instances: Poutsma, Grammar II. 1.A362, Bain I 193, Bog- 
holm, Bet. 134 — 135, Feldkeller 283, JSIyrop IV 204, Wellander, Lag- 
bundenhet 27, Paul, Prinz. 338, 343). 

The question for me is whether these "transferred epithets" as Bain 
calls them, show a change of basic meaning, or merely a fluctuation. 
Some instances clearly show the latter only. In other cases we should 
perhaps assume a real transfer. Thus with light, when transferred 
from a burden to the person carrying the burden: peo men Pet . . . god' 
untrussed lihte ase pilegrimes touward heouene ('the men that go un- 
burdened light as pilgrims toward heaven', Ancren Riwle, quoted from 
Stern, Swift 56). 

Transfers of this kind may be made for aesthetic reasons, and are 
then to be classed as metaphors: Mir wird's so wohl in deinem Ann, 
so frei, so hingegeben warm (Goethe, Faust); des Lichts gesellige Flamme 
(Schiller; both quots. from Feldkeller). Similarly in English: a rest- 
less pillow (Bain) . 


13.1. Theoretical Discussion. 

13.11. Analysis. The factor disturbing the balance of the meaning- 
complex in this type of sense-change, as well as in the following, is a 
modification of the subjective apprehension of the referent. In the 
present class we have a shift in the point of view concerning a detail of 
a total situation, a detail of a phrase referent, the same word being 
retained to denote it. 

The meaning of the English word boon was originally abstract, 
'prayer, petition, request', but later the word acquired a concrete sense, 
'thing asked for'. The shift is illustrated by sentences like he yatte 
hir freli al hir hone (1300), 'he gave her freely all her boon'. It is 
evident that the same total situation, the same phrase referent, may 
be adequately denoted by these words, whether boon is interpreted in 
its earlier, abstract, or in its later, concrete, sense. The meaning of 
the phrase changes, but the changed meaning is equally applicable to 
the referent. Other phrases with a similarly equivocal meaning are 
to ask a boon, to have one's boon, "taken without analysis" (ISTED). In 
all these cases, one detail of the -phrase referent may be apprehended, 
primarily, as the content of a petition, or, secondarily, as the thing 
asked for. The phrase referent is not changed, only another charac- 
teristic {Merkmal) becomes the object of attention and apprehension; 
the phrase remains an adequate description of its referent. 

Prayer and its object — the thing asked for — are correlative notions, 
and the thoughts of speaker and hearer are naturally carried forward 
from the prayer to its object, since the latter is the purpose and end 
in view (cf. Marty, Unt. 521, Jespersen, Language 175). When speaking 
of prayers being made or granted, the notion of the thing prayed for 
is therefore liable to form an element of the mental context. This makes 
for a very close association between the word with its primary mean- 
ing, on the one hand, and the correlative notion on the other hand. 

352 GUSTAF STERN 13-11 

By repetition the secondary notion, 'thing asked for', is so firmly 
associated with the original meaning, that the 'thing asked for' becomes 
a new characteristic [Merkmal) of the state of things described in phrases 
like to ask a boon, to have one's boon, etc. Under favourable circum- 
stances, the subject's attention may be especially attracted to the new 
characteristic, which then becomes a central element, while the pre- 
viously predominant elements of meaning are not attended to, and 
tend to disappear. God. . . sende the thyn bone (1385, Chaucer), 'God 
send thee what thou art praying for'.^) 

The analysis shows two points of essential importance, (i) One 
detail of the phrase referent may be apprehended in two different ways, 
while the phrase still remains an adequate description of its referent; 
the two notions being, in the extant circumstances, interchangeable. 
The detail in question serves as referent to one of the words in the 
phrase, and by the changed apprehension this word receives a changed 
meaning. (2) The primary and secondary apprehensions of the 
word referent are correlative notions, so that the latter is liable to form 
an element of the mental context of the former. 

Another instance is the word head, MB bede, originally signifying 
'prayer' (to God, the Saints, etc.). In the Middle Ages, prayers were 
mainly Pater Noster and Ave Maria, which had to be said repeatedly, 
and which were counted by means of the balls of a rosary. If it was 
said of a man that he was counting (or telling) his heads, the phrase 
signified, at first, that 'he was counting his prayers'. But a speaker 
looking on and using the phrase, would in reality see the man counting 
the balls of his rosary, and that action wotdd therefore be an element 
of the speaker's perceptual context. An association would thus be set 
up between the word beads and its primary meaning, on the one hand, 
and the idea of 'balls' on the other hand. It is also evident that whether 
the phrase in question is interpreted in one way or the other, it is an 
adequate description of the phrase referent (the action of the man) , 
which may be apprehended in either way. 

The first conclusion noted with regard to boon is applicable also to 

1) By a further change which is perhaps an adequation — the process is not 
quite clearly discernible from the quotations in the NED — the meaning of the 
word shifts to 'favour, gift, thing freely or graciously bestowed . . . without the 
notion of asking'. — Note that prayer, petition, and request all acquire the sense 
'thing asked for', but do not join in the further development of boon. 


beads. The second conclusion has to be somewhat modified. 'Balls 
of a rosary' and 'prayers' are not correlative notions, but in the given 
circumstances the former is an element of the (perceptual) context 
accompanying the latter. 

A shift in the apprehension of a complex referent, denoted by a 
phrase, will in most cases lead only to another word being employed; 
it might lead, for instance, to the phrase, he is counting the halls of his 
rosary. But in certain circumstances, as indicated above, the new 
apprehension of the word referent will, by repeated use, become asso- 
ciated to the word expressing the earlier apprehension of it, and will 
itself finally become a meaning of that name. A period of "preparation' ' 
thus seems necessary for a permutation. 

The process is, by definition, unintentional. If a speaker is attend- 
ing to his spreech he will not thus employ a word in a meaning that 
is not its own, and that is sometimes very different from its own. 

We may formvilate the process of change in the following manner, 
(i) A word is used in a phrase where a notion in some way connected 
with its meaning is liable to form an element of the context. (2) By 
frequent use the associated notion is associated also to the word. 
(3) The associated notion takes the place of the original meaning, in 
phrases of the type mentioned. (4) The word is used in the new, secondary, 
meaning, in phrases of other kinds, where the primary meaning is not 
possible (cf. Stocklein's analysis, p. 14). 

The process is too complicated to be adequately represented by a 
diagram, but the following scheme may perhaps serve to elucidate the 

Word: I. beads II. {balls of rosary) 

Meaning: a 

Referent: ^'T'" f. 

ab c d e f 

We may assume that the element y, belonging to the original mean- 
ing of beads, is 'prayers as being reckoned by the balls of a rosary'; 
while 6, belonging to the meaning of balls of rosary, is 'balls of rosary as 

Goteb. Hogsk. Arsskr. XXXVIII : i 23 

354 GUSTAF STERN 1^3-11 

being used to reckon prayers'. These two elements will naturally be closely 
associated — as indicated by the dotted line — so that we might 
even describe the notion of prayers as a characteristic of the 'balls of a 
rosary'. A person speaking of balls, in their aspect of being used to 
reckon prayers, that is to say, with the characteristic d for the moment 
predominant, will easily have his mind carried on to the associated no- 
tion y, and so to the word beads. By repetition the process may become 
habitual. This would not, however, by itself be sufficient to permit 
the emplojrment of beads as the name of 'balls', if it were not for the 
"functional synonymy" described in the next paragraph. 

The essential difference between this diagram and that given in 12.11, 
is the fact that in the latter we have a point of similarity in the two 
meanings involved {y and y^), while in the present case there is no such 
similarity {y and d are not similar, only closely associated). 

For the hearer, different alternatives might be imagined. He, too, 
might be looking at the actual counting of balls, apprehending it as 
such. Hearing the word counting beads, he would apply them to the 
counting of balls, and, if there had been a sufficient period of prepara- 
tion, he would not feel any discrepancy. On the other hand, if the 
hearer is not seeing the referent, but is only being told of it, the process 
for him starts from the word counting beads, and I will not venture 
to say if the notion of balls could be evoked, to the exclusion of the 
notion of prayers, before the new association had already been brought 
about in other ways. 

The basis of this class, as of the previous one, is to be found in the 
fact of specialization: the fact that referents have more than one charac- 
teristic, and that sometimes one, sometimes another of these charac- 
teristics is the object of attention (cf. 4.24). But there is an essential 
difference: in transfers, it is a case of two referents with a point of simi- 
larity (an identical characteristic) , which serves as a pivot for the trans- 
fer, making it possible to apply the name of one referent to the other. 
In permutations and adequations, we have one referent, with several 
characteristics. Through a change of apprehension, the subject's 
attention is shifted from one characteristic to another, so that the latter 
becomes predominant (central), and the meaning of the name is thus 
changed. In permutations, the referent is complex — a phrase re- 
ferent — and the change of apprehension concerns one detail of it; 
because this detail of the phrase referent constitutes by itself a word 


referent, the meaning of its name, of the relevant word, is changed. 
In the next class — adequation — we shall find yet a third type of 
change based on specialization. 

If we make the assumption that the interchangeability or functional 
synonymy described in the next paragraph mediates the shift, we may 
formulate an alternative explanation of the psychic process. Refer- 
ents which are not similar, but otherwise related can have no com- 
mon element within the habitual range of their names; but the names 
may in certain circumstances be functional synonyms. The capability 
of functioning in a certain position in a phrase might be regarded as 
an element of their meaning for the moment. They have consequently 
a temporary common element of meaning, which might mediate the 
shift in the same way as the common element in a transfer. The dia- 
gram for a permutation would then be identical with that for a transfer, 
although the point of similarity would be of a peculiar nature, not 
permanently belonging to the semantic range. 

This theory, too, explains all unintentional sense-changes (except 
those caused by substitutions, analogy, or shortening) with the help of 
two fundamental principles, but the permutations are bracketed with 
transfers as showing a shift mediated by a point of similarity in two 
referents, instead of being classed with adequations as illustrating a 
shift of attention from one characteristic of a referent to another. For 
my part, I consider the latter theory as more likely to be the true 

Another point that may be mentioned in this connection is the distinc- 
tion between sense-changes based on association of similarity and those 
based on association of contiguity, which appears in more than one 
classification, and which corresponds more or less to Wundt's distinc- 
tion of shifts with changing and with constant predominant elements 
(Wundt II 531). The recurrence of this dichotomy would seem to 
indicate that several writers have felt it to be indicated by the actual 
circumstances; the question is how it should be explained. 

The whole problem is more psychological than linguistic, and I have 
to leave it with these remarks. 

13.12. Conditions. The Equivocation. One basic condition of 
permutation has already been indicated: a word must appear in such 
a context that a notion forming a constant, or at least a frequent, 
element of the context is associated with it. 

356 GUSTAF STERN 13-12 

But it is not enough that the secondary notion forms an element of 
the context of the word as used with its primary meaning. The two 
notions, the two meanings, must be interchangeable, or more precisely, 
functional synonyms; that is to say, either of the two word-meanings 
may be inserted into the ^^r«se-meaning, and the latter still remain an 
adequate apprehension of the phrase referent. There are many notions 
associated with the notion of prayer, but in the phrase he is telling his 
heads, it is only the notion 'ball of rosary' that can be substituted for 
the notion of 'prayer' without making the phrase inapplicable to the 
phrase referent. It is this characteristic of the secondary notion ('ball 
of rosary') that enables it to take the place of the primary meaning 

There must, consequently, be some sort of connection between the 
two notions that constitute the primary and secondary meanings of 
the word. The nature of this connection may vary within very wide 
limits. In the case of boon, it is a logical correlation; in the case of 
beads it is a fortuitous, temporary, connection, founded on the practice 
of using rosaries for counting prayers. Other permutations show 
other kinds of connections, as will be evident from the instances quoted 
below. The nature of the connection would perhaps repay a closer 

Taking the matter of functional synonymy from the point of view 
of the word itself, we may say that the meaning of the word is equi- 
vocal. I distinguish equivocal, in this special sense, from ambiguous, 
which means that the phrase may refer to either of two different re- 

The term equivocation should not be interpreted as signifying that a 
speaker vacillates between two meanings, still less that he is aware of an 
alternative. Permutations are unintentional. "Equivocal" must therefore 
be taken to mean that if the speaker turned his attention to the matter , 
either of the two possible meanings might emerge as the one really 
intended. This lack of precision, on the other hand, is possible owing 
to the fact that we are able to apprehend a total with clearness, without 
clearly apprehending every detail of it. (Cf. 4.17). The word, 

1) I have previously (Stern, Swift 11 sqq.) employed the term oscillating, but 
since there is no movement forward and back, even in a metaphorical sense, I 
now prefer to speak of equivocation. Marty employs the term equivocation in a 
different way. 


for instance heads, must be assumed to carry a sufficiently definite 
reference to that which is being counted, but since either way of 
apprehending this referent may function in building up the total 
meaning (the phrase meaning), we leave it at that. We know what 
the word means, and need not trouble to reflect how it means. When 
the "Washestimmtheit" is sufficiently definite, the "Wiebestimmtheit" 
may be left vague. The meaning of the single word is in some way 
"lost" in the meaning of the phrase. The precise nature of the 
psychic processes is still to be ascertained. 

A further condition for permutations is, as already stated, a "period 
of preparation" during which phrases with the equivocal meaning 
occur with sufficient frequency to allow a firm association to be estab- 
lished between the word (with its primary meaning) and the notion that 
comes to form its secondary meaning. In order to be convincing, an 
explanation of a sense-change through permutation must be illustrated 
by phrases — quoted from texts — in which the relevant word is na- 
turally and frequently employed. When the association in question 
has been established, the way is open for the final stage of the shift: 
the use of the word in the new meaning without equivocation, and 
without implication of the primary meaning. 

My argumentation is throughout based on the assumption that if, 
in an unintentional change of meaning, the relation between the prim- 
ary and the secondary meaning is not one of similarity, and the 
shift is not an adequation (see next chapter), then the shift cannot take 
place — unintentionally — except under the conditions indicated above 
for permutations: the two meanings must be apprehended as alter- 
native characteristics of a given total (the phrase referent) and they 
must be functional synonyms in the given verbal context. Intention- 
ally such a shift may, at least in many cases, be effected without the 
mediation of any specific context (cf. 11. 7). 

I assume, consequently, that the step from one meaning to the other, 
when they have no identical element, is too great to be effected unin- 
tentionally except when the circumstances are especially favourable. 
The nature of these circumstances has been described above. It fol- 
lows, as a methodical rule, that we should always seek to ascertain in 
what verbal contexts the word in question was used prior to the emer- 
gence of the new meaning. These contexts will show whether we have 
really a permutation, or a shift of another kind. 

358 GUSTAF STERN 13- 1 2 

Something must be said on the nature of the relation between the 
two referents involved in a permutation. In the case of boon, they 
are logical correlates; in the case of heads, the connection between 
them is extrinsic and temporary. In some instances the connection is 
still more "singular", if I may use Wundt's term, which is really appli- 
cable here; the two referents may have nothing to do with each other 
except in one single context. An instance in point is afforded by the 
word premises. 

The premises originally meant 'the aforesaid' . It was often employed 
in announcements and official documents concerning the selling or 
letting of houses and lands, where at first all the details were enumera- 
ted: 'the house with outhouses, garden, land, tenements' etc.; further 
on in the document, the writer inserted instead the premises, 'the afore- 
said'. The referent of premises is, on the one hand, 'previously men- 
tioned', on the other hand, it is a 'house with outhouses, etc' Either 
way of apprehending the referent may function in the given context: 
the two notions are functional synonyms. The secondary notion there- 
fore becomes associated to the word premises, and finally supplants 
its earlier meaning. 

The "step" from 'aforesaid' to 'house, lands, or tenements (above- 
said or before-mentioned)' is, from a logical point of view, very long, 
and this circumstance has led some writers to assume that shifts of 
this kind are due to misunderstanding on the part of readers or hearers. 
But there is no difference in kind, only one of degree, between the 
process of change in boon and that in premises, and there is consequently 
no reason for seeking a fresh explanation for the latter. Whatever 
the logical connection between the referents, there is always a very 
close factual and psychological connection, which allows us to regard 
the shift as a normal phase of linguistic development. We should 
therefore explain it by normal psychic processes, liable to take place 
in the mind of any person, ignorant or not. It is contrary to sound 
methodical rules to seek the explanation in misunderstanding, which 
is an abnormal factor (cf. 7.42).^) 

^) "Permutations by mistake" do sometimes occur, when a word for some 
reason is misunderstood, and grows habitual in the incorrect interpretation. 
Derring-do is a curious instance. The NED describes the process in the following 
way: "The two words durring, dorryng, daring, vbl. sb., from durran dorren to 
dar©, and don do, pres. inf. of do, v., literally daring to do, which, by a chain of 


13.13. Causes. With regard to the causes of permutation I assume 
that, generally speaking, they are to be sought in the striving to fulfil 
as adequately as possible the symbolic and communicative functions 
of speech, but it is difficult or even impossible to state the cause with 
any precision in the individual instances. The diagram in 13. 11, 
together with the explanations given in connection with it, show only 
the possibility of permutation, not the causa movens. We have to 
ask why, for the speaker, the word beads offers itself and is made use 
of, although the referent spoken of is 'balls of rosary'. 

In the case of premises, that word is chosen because it is so much 
shorter and more convenient than the enumeration of sundry details 
concerning the house etc. But in most cases, economy of effort cannot 
be adduced. Cf. the remarks on beads in 13. 11. Similarly with 
regard to boon. Various factors might be suggested, but we cannot 

misunderstandings and errors, have come to be treated as a kind of substantive 
combination, taken to mean, Daring action or feats, "desperate courage". The 
words come incidentally in their ordinary sense and construction followed by the 
object 'that' ( = what, that which) in Chaucer's Troylus; whence, in an imitative 
passage by Lydgate, in an absolute construction more liable to misunderstanding; 
Lydgate's dorryng do was misprinted in the i6th c. editions derrynge do, in which 
form it was picked up by Spenser and misconstrued as a subst. phrase, explained 
in the Glossary to the Sheph. Cal. as 'manhood and chevalrie'. Modern roman- 
tic writers, led by Sir W. Scott, have taken it from Spenser, printed it derring-do, 
and accentuated the erroneous use". Chaucer's lines in Troylus (V 837) run: 
Troylus was neuere vn-to no wight ... in no degre secounde. In dorryng don pat 
longeth to a knyght {i. e., in daring to do all that becomes a knight) . . . His herte 
ay wih he firste and wih he beste Stod paregal, to dorre don that hym teste. Lyd- 
gate writes of Troilus (Chron. Troy. II xvi): And parygal, of manhode and of dede, 
he was to any hat I can of rede, In dorryng do, this noble wor^y wyght, Ffor to ful- 
fille pat longep to a kny^t. 

Another instance is warison, originally 'possessions, wealth, reward', which 
was misunderstood by Sir W. Scott, probably, according to the NED, in a phrase 
from The Battle of Otterbourne: Mynstrells, playe vp for your wary son. Scott inter- 
preted this as 'note of assault'. Byron and others have taken the word from Scott. 

But the explanation "by mistake" should not be extended beyond very definite 
limits. Thus, I think it is not correct when the NED states, of the word transpire, 
originally 'to escape from secrecy to notice, to become known, to leak out', later 
on 'to occur, to happen, take place': "Evidently arising from misunderstanding 
such a sentence as what had transpired during his absence he did not know" . The 
change is a permutation of the ordinary type, with equivocal phrases. If we assume 
it to be due to mistakes, then language is fuU of them, and most of the instances 
in this chapter should be so called. 

360 GUSTAF STERN 13-13 

fix on any of them as clearly responsible for the shift. Possibly we 
should assume an intermediate stage, in which a speaker, wishing to 
denote the referent 'gift', apprehends it as being asked for, and this 
subsidiary characteristic may evoke the word boon. Possibly we should 
assume, at least in some cases, that a given phrase is habitual in a 
specific situation, and is used even if one detail of the total is differ- 
ently apprehended: that may be the case in ask a boon, have one's 
boon, and so on. And it is evidently the case with premises, and to 
count one's beads. 

Some writers explain permutations as due to a difference between 
the meaning intended by the speaker and that comprehended by the 
hearer; thus lycumann, IF 45. Such a difference is of course possible, 
and no doubt occurs pretty frequently, but I do not think that 
we should make it the basis of a general explanation. The hearer, like 
the speaker, must be assumed to know his native language; permuta- 
tions are too common to be due to misunderstandings. I have insisted 
above that to explain them in this way is an incorrect method, because 
they are a normal phase of linguistic development, and should therefore 
be explained by normal psychic processes. 

For the hearer, the case is somewhat different. I refer to the re- 
marks in 13. II, in connection with the diagram. 

13.14. Delimitation and Definition. The previous argumentation 
is founded on the assumption that if the relation between the two 
referents involved in an unintentional sense-change is not one of simil- 
arity, as in transfers, and if the change is not an adequation, the pro- 
cess of change follows a peculiar course, as described. Consequently, 
I distinguish permutations from regular transfers by the nature of the 
relation between the two referents. 

From adequations they are distinguished by involving the mediation 
of a specific context, in which one detail of a phrase referent is differ- 
ently apprehended. The adequation is a matter of a single word. 

Permutations merge into popular metaphors founded on other rela- 
tions than similarity, and the criterion of intentional and uninten- 
tional is, as usual, frequently difficult to apply in the individual in- 
stances. It has been discussed at length in 11. 13. 

Permutations also merge into mere fluctuations; this distinction 
must be left to future research to determine more precisely. 

These arguments result in the following definition of permutations: 


permutations are unintentional sense-changes in which the subjective 
apprehension of a detail — denoted by a separate word — in a larger 
total changes, and the changed apprehension (the changed notion) is 
substituted for the previous meaning of the word. 

13.15. Classification. As in the two preceding classes, I distinguish 
first the parts of speech. It is only of nouns that I have a sufficient 
number of instances to make further distinctions. The relations be- 
tween the referents are of many different kinds, but not all are frequent 
enough to constitute the basis of a considerable number of changes. 
The instances given below represent only a selection of the possibilities. 
I have generally quoted one or more equivocal instances for each word, 
but I have not thought it necessary to give further quotations, since 
the words are all of them well-known, and may easily be looked up in 
the N:ED. 

It will perhaps be possible at some future time to classify the permu- 
tations according to the relations between the two referents involved, 
but at present such a procedure would hardly lead to satisfactory 
results. I have therefore adopted a logical, and practical, arrangement. 

Some of the shifts exemphfied in this chapter could no doubt be 
effected intentionally, or through shortening, or analogy, and would 
then have to be classed accordingly. Some of my explanations are 
therefore tentative only, but when equivocal instances are found, 
from the period when the new meaning emerges, I have considered 
permutation as a probable, or at least a possible, explanation. 

The arrangement below is, as far as possible, parallel to that of the 
preceding chapters. 

The circumstance that a great number of sense-changes are medi- 
ated by equivocal phrases is so conspicuous that it has naturally attrac- 
ted the attention of many previous writers. I may mention Hecht, 
Hey, Stocklein, Collin. But no one has, as far as I know, attempted 
to ascertain the precise nature of the mediation, or the extent of the 
type and its connections with and differences from other types. My 
own theory and the definitions built on it are therefore a first attempt 
to solve the problem, and I offer my explanations with due reservations, 
since the material at my disposal is on many points inadequate. When, 
as is often the case, the same semantic result may be attained either 
by permutation, nomination, or shortening, I am sometimes unable to 
state with confidence what the actual development has been. Further 

362 GUSTAF STERN 13-15 

research in these matters will perhaps compel us to draw the lines 

13.2. Nouns. 

13.21. Objects' Names {Concrete and Abstract). 

13.211. Material for Object made from it. The type is old. We 
find in OE. iren used for iron articles; cf. Latin ferrum, aiirum, aes, etc., 
similarly applied. The process in these old instances cannot be ascer- 
tained. In other cases the change is later, and equivocal instances 
are found in the NED. 

Brass 'a sepulchral tablet of brass': Payd for fasting of the brass of 
the graves in the chaunsells vd. (1613). Also, 'a musical instrument of 
brass': as bras sownynge or a symbal tynkynge (1382). This might be 
a sense-loan from Latin. Also, 'brass or copper coins': Beere heor 
bras on J>i Bac to Caleys to sylle (1362). Posses not golde, nor silver, nor 
brasse yn yoiire gerdels (1526). — Copper, 'copper money': // so, our 
Copper biiyes no better treasure (1588). — Lead: 'a bob or lump of lead 
suspended by a string to ascertain the depth of water': / sail 
caste leede and loke J>e space (1440). Note the absence of the de- 
finite article, which shows leede to be equivocal in meaning. — Ma- 
hogany, 'a table, esp. a dining-table': / had hoped . . . to have seen you 
three gentlemen. . . with your legs under the mahogany in my humble 
parlour (1840). Perhaps a metaphor? — Sacking 'a material used for 
making sacks' > 'a piece of such material': His Horses stand with Sack- 
ings instead of Cloaths (1707). — Sackcloth, 'garments of sackcloth': 
he sits him down in sackcloathes, his hands and eyes reared to heauen 

With regard to the use of brass as denoting articles made of brass, 
the NED says that it is elliptical. This may very well be the true 
explanation in some cases, just as beaver, for a beaver hat, is probably 
elliptical, if the dates in the NED are reliable. But in the instances 
quoted above the presence of equivocal phrases shows that permuta- 
tion is a possible explanation. 

In other cases, the use of a material for an object is clearly metaphor- 
ical, as when a sword is termed in Beowulf dryhtlic iren 'a splendid 
iron', or when a ME text speaks of Laurear of martirs, foundid on 


holynes (1430), or when a diploma, in U. S. university slang is called a 
sheepskin^) (Cf. 11.73). 

13.213. Receptacle for Content. Barrel is used for 'the contents 
of a barrel': For they no haveth joye . . . Bote in the gutte, and the bar ell 
(1300). A barrel of water is equivocal, and its meaning may be spe- 
cialized either on the receptacle or on its content. — A tub is often 
used for the contents of a tub: 'the water for bathing': He was in his 
tub (1861). An aversion for the Saturday tub (1849). ^^ ^^^^ ^ ^old 
tub. — Box may be 'the money contained in such a box [i. e., contain- 
ing public funds), a fund for a particular purpose': He schal haue of 
f>e comune box xiijd (1389). — Sack sometimes denotes the contents 
of a sack, or the amount usually contained in a sack: LI saks & X 
peres de leine (1314). For iij sak lyme to pe same mason . . . vj d. (1427). 
And similarly barrow, basin, beaker, bowl, glass, etc. — A wardrobe was 
first 'a room in which wearing apparel was kept, a dressing-room', 
later,' a person's stock of wearing apparel': Jupiter hath in his warder obe 
bothe garmentes of joye and of sorowe (1387 — 88). - — • Post was originally 
'one who travels express with letters, messages, etc.; esp. on a fixed 
route; a courier, a postrider, a letter-carrier, a post-man', and also 
'a vehicle used in the conveyance of the mails; a mail-coach or -cart; 
a packetboat'. From these meanings arose a new sense: 'a single 
dispatch of letters (and other postal matter) from or to a place; also 
concretely, the letters, etc., collectively, as dispatched or conveyed, 
with that which carries them, the mail'. The NHD points out that 
this sense is often blended with the two preceding, as in The post being 
just going, I can say no more (1675), by the last night's post (1683), upon 
the coming in of the post (1711). Here the word post, originally signi- 
fying the conveyance or the person carrying the mails, would easily 
come to be associated with the thought of the mails themselves, these 
being the matter of interest both for writer and reader. 

Similar shifts may be made metaphorically for stylistic purposes, 
as in the power of the purse (cf. Nyrop IV 198). 

13.213. Part or Constituent Detail for the Whole, and Vice Versa. 

1) With regard to the instances quoted by Nyrop IV 201 and Carnoy 179, it 
is often doubtful which of the three alternative explanations is the right one. 
The following, however, appears to belong to the present group: De jeunes filles . . . 
dans leurs grisettes, the last word signifying originally a cheap grey stuff, and 
later on, a dress made from this stuff. 

364 GUSTAF STERN 13-213 

Pars pro toto is a frequent heading in semantic classification, but is 
not a uniform type; the same result can be attained in more than one 

Hand 'a person employed in any manual work, a workman or work- 
woman'. Those hands that must he employed in their building (1657). 
Two hundred manufacturers, which would employ many hands (1721). 
Come, gentlemen, all hands to work! (1703). Additions by the hand that 
retouched the writing (1893). — Blade is often put for sword, as being 
the essential part of the weapon: A long surcote of pers vp on he hade, 
And by his syde he baar a rusty blade (1386, Chaucer). A shift of this 
kind may also be metaphorical. — Hearse 'a carriage or car constructed 
for carrying the coffin at a funeral', is originally the superstructure of 
such a vehicle: 'a frame-work of wood used to support the pall over 
the body at funerals'. It fitted on to the parish bier, and was probably 
adapted to carry lighted tapers. It is easily conceivable that in speak- 
ing of the hearse, in the earlier sense, the mind was carried on to the 
thought of the whole vehicle. 

F. bureau 'writing-table' (originally the piece of cloth covering the 
top of the table) is also taken to mean the whole office in which the 
table stands. "Quand on est a son bureau, on est aussi, par necessite, 
dans son cabinet de travail" (Nyrop IV 198. Many of the instances 
adduced by Nyrop seem to be metaphorical, as I take that term. Cf . 
also estaminet and toilette. Carnoy 158, Weekley, Words 73, and Bain 
I 191, with instances of so-called synecdoche). 

Instances of totum pro parte are doubtful; they are more often ade- 
quations. It is mostly a question of one particular aspect of the total 
referent, which becomes more or less habitual, and finally emerges as 
an independent meaning of the word. A dungeon is originally a strong 
tower, the central tower of a fortress. Since such towers were generally 
furnished with cellars used as prisons, to put a man in the dungeon woiild 
mean putting him in prison, and this meaning of dungeon gradually 
became the most common one; in the sense 'tower' the spelling donjon 
is now mostly used. Tower is also employed of such structures used 
'as prisons' (NED): Se biscop Rannulf . . . ut of ^am ture on Lunden 
nihtes odbcerst (1122). The names of trees are used to denote the wood 
or the leaves or other parts of them: Do per to sage and persely soyng 
(1420: the leaves of sages and parsley). Smal-coale. . . is made of Sallow, 
Willow, Alder, Hasell, and the like (1646: the wood of these trees). 


I am uncertain whether such instances should be classed as per- 
mutations or adequations (cf. 14.83). A detailed knowledge of their 
history would be necessary to decide the question. 

13.314. Symbol for Thing Symbolized. This shift is often metaphor- 
ical (11.72). It seems to be a permutation mainly when the word is 

A bush is 'a branch or bunch of ivy hung up as a vintner's sign; the 
sign-board of a tavern', also 'the tavern itself: Twenty to one you find 
him at the bush (1625. Cf. Nyrop IV 196). — Brand is 'trade-mark', 
but also 'a particular sort or class of goods': The ale was of a superior 
brand (1854). 

Here we may perhaps class wedlock, originally 'the marriage vow or 
obligation'. Chiefly in phrases, to hold, keep wedlock 'to be faithful in 
marriage', to break [one's) wedlock 'to commit adultery'. And tohh 
wass heh & sop weddlac Haldenn onn ess^er hallfe (1200). dif ha hare 
mediae laheliche halden (c. 1320). In such phrases and combinations 
the word could be interpreted as meaning 'the condition of being mar- 
ried, marriage as a state of life or as an institution, matrimonial rela- 

13.215. Instrument for Action. I have only one good instance. 
Ballot is 'a voting paper', later 'the act of voting': A triall of theyr 
sentences by Ballot (1549). 

13.216. Instrument for Product. Whistle, 'instrument' > 'sound 
of whistle': Whan Marcuryis whystyl hym dede streyne To hys deed slepe 
(1447). The sound of pastoral reed . . . Or whistle from the Lodge (1634). 
Lucky that Klepper knows my whistle, and follows me as truly as a hound 

The NED suggests that the sense 'sound of whistle' is a new forma- 
tion from the verb, but the equivocal instances adduced show that 
permutation is also possible. Cf. a similar development in F. sifflet, 
according to Nyrop (IV 206, with comments), and E. pipe, at first, in 
OE, a wind instrument, later (earliest inst. 1580) 'the voice, esp. as 
used in singing, the song or note of a bird, etc' The following instances 
seem to be equivocal: Where vnder a sweete Arbour . . . be byrdes record- 
ing theyr sweete notes, hee also strayned his olde pype (1580). A straunge 
orator straining his pipes, to perswade straunge people (1581). Thy small 
pipe Is as the maidens organ, shrill, and sound (1601). 

13.217. Organ for Capability of Perception or Intellection. Ear is 

366 GUSTAF STERN 13-217 

used in sing, and plur. for 'the sense of hearing, auditory perception'. 
To come to the ear{s of: to come to a person's knowledge by hearing 
(NED). It com the kinge to ere (1297). The tithandis . . . Com to the 
cliff urdis ere (1375). — Eye is used in sing, and plur. for 'the action or 
function of the eyes, the sense of seeing'. Chiefly in phrases, {To have) 
before one's eyes, lit. and fig., To believe one's {own) eyes, To catch, fix, 
strike, take the eye (NBD). He litlede him seluen to-foren mannes ei^en 
(a. 1200). Pal for a tym desceyui}. . . pe 3ee, but }is biggip pe vnderstonding 
perpetual (1400). — Nose is similarly used of the sense of smell: Quare- 
of J)e breth as of bawme blawis in oure noose (1400 — 50). Was dulcet & 
swete to y^ mouth. . . & sauoured wele to the nose (1526). — T^he palate 
is "popularly considered as the seat of taste", and the word was there- 
fore transferred to the sense of taste: Breed to a sore mouth is sharpe & 
harde, whiche to a hole palate is swete & pleasaunt (1526). Let their pallats 
Be season d with such Viands (1596, Shakespeare). Tooth is similarly 
used, 'referring to eating, esp. to the sense of taste': / wol kepe it for 
youre oinene tooth (1386, Chaucer). The NED explains this as figura- 
tive, which may be right; a similar explanation is not impossible for 
the other instances quoted above. Foreign influence is also possible. — 
Brain is used for 'intellectual power, intellect': That is nought for lake 
of braine (1393). 

Object for Perception. According to Wundt it is a general rule 
that perceptions are named from exterior objects or states connec- 
ted with them. This is due to the fact that no perception is independent 
of such objects or states. We do not perceive blue colour, or a note, 
as pure perception, but as a quality or an activity of some object out- 
side us. Most modern colour names that are not arbitrary (as Prussian 
blue), are the names of objects or materials: orange, terracotta, indigo. 
(Early names have been investigated by Bechtel, with similar conclu- 
sions. Cf. also Falk 14, Wundt II 556). 

I have only one English instance that seems to belong here. Sight 
is originally, in OE, 'a thing seen, esp. of a striking or remarkable 
nature, a spectacle'. In early ME, it comes to be used also of 'the per- 
ception or apprehension of something by means of the eyes; the pre- 
sentation of a thing to the sense of vision'. Equivocal instances are: 
se schulen hebben . . . pe brihte sihde of Codes nebscheft (1225). Pis leuedi 
duted noght pe sight pis angel pat was sa bright (1300). Wei hath For- 
tune y-turned thee the dys. That hast the sighte of hire, and I thabsence 


(1386). (He) was as deed only by the syghte of the sayde dragon 


13.218. Articles of Dress or Equipment for Person. The earliest 
instances of this type are some names of military flags, weapons, and 
musical instruments, used to denote the person carrying or using them 
(see A. Smith, who gives a full material with copious quotations). The 
type seems to be an imitation of French usage in military language 
(see Nyrop IV 195, with some early instances), and the practice after- 
wards spread to other articles of dress, tools, etc. From the 14th c. we 
have standard 'standard-bearer'. The NED has no early equivocal 
instance, but cf.: Euery man mounted, and the haners and slanders 
folowed this new made knyght (1523). NED interprets this literally, but 
evidently the secondary meaning is also possible, as well as another 
secondary meaning, 'company of cavalry'. From the 15th c. we have 
helm: The kyng of Northgalys with eyght score helmes (1470-85), and 
trumpet: The trompettis vppon the wallis went (ai45o). Cf. the earlier 
Latin quot.: Dati a le Trumpet de dono domini ibidem . . . (1390 — 91). 
The Duke of Brunswicke sendeth a trompet to Duke Moris and desyreth 
a communication (1560). 

In the i6th c. the practice grows still more frequent. We have in- 
stances of bow, corslet; drum, fife, rebeck, clavicord, shawm, sackbut, 
bag-pipe; banner, ensign, guidon, pensel, all denoting the persons using 
them; the last four also denote bodies of men. This seems to be an 
international practice, cf. German Fdhnlein, Fahne, Swed. fana, fdnika. 
The NED does not give early equivocal instances of all these words, 
but the following seem to be worth quoting: 

The lord lieutenant . . . sent a drum vnto Monsieur Doisell (1577 — 87). 
There was among these a thirtie bowes with a bagpipe (1577). Esquires . . . 
able at the Musters to present a Launce or light horse for the Prince's 
sendee (1602). To Doctre Lee's shawmes and shagboshcs [sackbuts) that 
playt before my Lorde of Solfolke . . . (1539). There sat dame Musyke, 
with all her mynstrasy ; Rebeckes, clarycordes, eche in theyr degre, Dyd 
sytte aboute theyr ladyes mageste (1509). All the companyons, to thenombre 

1) Wundt (II 561) classes along with this group also "die Uebertragung von 
Benennungen ausserer Eindriicke auf subjektive Gemiitszustande und auf die 
psychischen Krafte von denen diese abhangig gedacht werden ... So kann 
angustia sowohl die raumliche Enge wie das Gefvihl der Not . . . bedeuten". This 
is, in my classification, either a metaphor or a transfer, probably the former. 

368 GUSTAF STERN I3.218 

of xii. hundred pensels, And they were right hardy and valyant knightes 
(1523). For one monthes wages . . . for iiij drummes and two fyfes, 
every at xl. s. (1548). Item, for Thomas Evans, Reheke, wages xx s. (1540). 

In the i6th c. names of non-military articles were similarly used; 
there are instances of Chrisom, blue-coat, red-coat, coxcomb, red-hat, 
home-spun, rochet, clout-shoe; and later on the usage spreads. 

As pointed out above (11.71), shifts of this kind may be metaphors; 
the earliest instances may be borrowings from the French, and later 
instances in English may be formed analogically. Shortening is perhaps 
not altogether out of the question in some cases. And finally there 
is the possibility of permutation, proved by the equivocal instances. 
In some of the popular formations, permutation seems to be indicated 
by the nature of the phrases in which the words are used, for instance: 
/ cannot work with a petticoat in the room. The matter is far from clear, 
and a detailed study would be welcome (cf. Bain I 191, Carnoy 261). 

With regard to the modern use of boots 'boy cleaning the boots', 
Jespersen (Grammar II 5. 723) assumes that the secondary meaning 
arose from the habit of calling boots! when people wanted their boots 
cleaned; the call was apprehended as a call for the boy or man who did 
the work, and so became his name. If this is correct, the change is a 

13.219. Name from Concomitant Circumstances. Since permuta- 
tions go by factual connections, not by logical relations, they cannot 
all be fitted into a logical scheme like that represented by the groups 
in the present chapter. I am therefore obliged to add a section to in- 
clude various instances in which we can only say that the naming is 
based on concomitant circumstances; the exact connection being of 
various kinds. 

Board is 'table used for meals' and changes to 'food served at table'. 
Equivocal instances are: Mi bord is maked. Cumed to borde (1200). 
Hwon gredie hundes stonded biuoren ^e borde (1225). Heo seten to borde 
(1275). The change is completed in: To pay for bord (1386). — Hiftt 
is originally 'occasion, opportunity', now 'a slight indication intended 
to be caught by the intelligent, a suggestion or implication conveyed 
in an indirect or covert manner': Vpon this hint I spake (1604). — 
Knot is the speed of a vessel measured by paying out a line, furnished 
with knots at every 50 feet; the number of knots running out in half 
a minute was equal to the number of nautical miles that the ship sailed 


in an hour. The speed was therefore indicated by saying that a ship 
was sailing five knots. Since this speed was equal to five nautical miles 
per hour, knot acquired the meaning of a unit of measurement, primarily 
the divisions of the log line. — Quarter 'region, district, place, locality' 
is originally the region lying about or under one of the four principal 
points of the compass or divisions of the horizon, and still earlier one 
of these points or divisions themselves. Vpon Elam I wil bringe the 
joure wyndes from y^ foure quarters of heauen (1535, Coverdale); here 
quarters may be interpreted as signifying 'the points of the compass' 
or 'the regions lying in that direction'. The latter meaning then was 
subjected to adequation, during which the connection with the points 
of the compass disappeared, and the sense 'region, district' became 
free from such implications. 

Score is originally 'a notch cut in a stick or tally, used to mark num- 
bers in. keeping accounts'; it was used equivocally in sentences where 
it might denote the tally itself: All that Launjal had horwyth before 
Gyfre, be tayle and be score, Yald hyt well and fyne (1460). Yf J>o koke 
wolde say pat were more, pat is po cause pat he hase it in skore (1460). 
The word further acquires the sense of 'a record or account kept by 
means of tallies': pe fendes redy my rolle to rede, pe countretayle (the 
opposite half of a tally) to shewe, pe score (142 1). Out of this sense 
there further develops another: 'the sum recorded to a customer's 
debit in a 'score': the amount of an innkeeper's bill or reckoning, etc.': 
After he scores, he neuer fayes the score (1601). Or how to pay thy hinds 
and clear all scores (1648). When in the morning Matt ask'd for the 
score, John kindly had paid it the evening before (1701). ISTED also 
gives the phrases to clear, pay, quit a score, and others. 

Fr. danger is from dominiarium, from dominus, and means originally 
'domination, power': Melons nos hors de lor dangier (Wace, Roman de 
Rou). This sense did not long survive the Renaissance. The trans- 
ition to the modern sense occurred in phrases like the following: Les 
cardinaux, qui se veoient au danger des Romains (Froissart): if the car- 
dinals were au danger 'in the power of the Romans, they might also be 
in danger from them. Cf . the phrase tirer quelqu'un du danger de mort 
(Nyrop IV 77). 

12.23. Nomina Actionis. 

13.221. Action for Product, Result, or Object. A very common type. 
Cf. Nyrop IV 221 sqq., Collin, Ata 54 sqq. 

Goteb. Hossk. Arsskr. XXXVIII: i. 


Batch 'the process of baking' > 'the quantity of bread produced at 
one baking': He shall truly e delyver into the bredehouse . . . the whole 
numhyr of his hache (1461). Similarly Swed. hak: vi ha tre hak att gora 
i dag (SAOB). — Lending 'act of lending' > 'something lent': Thou 
lost a good wife, thou lost a trew friend, ha? Two of the rarest lendings 
of the heauens (1602). — Schooling 'the maintenance of a child at school, 
considered as involving expense; hence, cost of school education': 
Find my Children Apparel and their Schooling (1610). She could not 
afford to pay for her little lass's schooling (1802). — Leading 'the act of 
leading' > 'that which is led': Thai that war of his leding . . .war all 
ded (1375). He hadde in his ledyng (Vulgate: in comitatu) chares, and 
rydynge men (1382). Al the folk of Mr leding . . . never wist what was 
fleing (1400). — Cast 'act of casting or throwing' > 'the distance which 
anything can be thrown': Filers as hi^ as a stones cast (1387). Also 'a 
throw or stroke of fortune; hence, fortune, chance, opportunity': Him 
sidd J>an reu his cast fat }is folk was fra him past (1300). Glaid of 
this cast, seand thair tyme maste gane (1513). — Frayer 'supplication' > 
'the matter of a petition, the thing prayed for or entreated': Thus hath 
he graunted my pray ere (C1400). Similarly petition and request. Cf. 
boon, 13. II. — Lift 'act of lifting' > 'the thing lifted': Off gold well 
twenty mennys lyffte (13 . . ) — Damage is 'injury, harm' and also (now 
always in the plural) 'the sum of money to be paid in compensation 
for loss or injury sustained'; the word could be taken both ways in 
a phrase like to pay damage. The party condemnyd . . . schold euer be 
awardyd to pay costys and al other dammage cumyng to hys aduersary 
by the reson of the vniust sute and vexatyon (1538). — Business 'state or 
quality of being busy' > 'that about which one is busy': Me to serue 
is al his besynesse (1392). He that wele & dyligently vnderstondith to his 
bysenesse (1477).^) 

1) The type is analysed at length by Leumann (IF 45, no). The use of Schreiben 
for 'Geschriebenes, Brief is ascribed to phrases like ich bin mil dem Schreiben (der 
Tatigkeit des Schreibens) zu Ende, "auf die der andere aus der Situation heraus 
schreiben als 'Brief deuten und also sagen konnte: dann kann ich das Schreiben 
gleich mitnehmen. Man ruft die Kinder kommt zwn essen, sie verstehen zum Essen 
(wie zur Milch) und sagen etwa, das Essen ist noch zu heiss." I have already stated 
that I do not believe in the general validity of the theory that the change is due 
to a difference of apprehension between speaker and hearer. Leumann commits 
the mistake of not adducing historical instances; he constructs phrases to explain 


13.222. Action for Instrument or Means of Action. Aid 'action of 
aiding' > 'something by which assistance is given, a means or material 
source of help': Exercise may deserve to be taken as a common Aid to 
Physick (1711). Neyther lakt I ayde in any wicked dede (1559). — Fare 
'a going, journeying, course' > 'a passage or excursion for which a price 
is paid', and hence 'cost of conveyance, passage money': Pare suld nane 
pay mare ^an foure pennys for pare fare (1425). He payde his fare and 
wente aborde (1535). Most willingly I'le pay thereof the fare (1620). 
Cf. German Fuhr 'das Fahren' > 'a cart': Eine Fuhr Heu 'a cart of 
hay' (Stocklein 58). Cf. Swed. farkost, Collin, Ata 88. Call is 'a par- 
ticular cry or sound used to attract or decoy birds, etc.', and also 'a 
small instrument or whistle to attract birds, etc. by imitating their 
note': The deer came . . . as if they had been used to a keepers call {1596) . 

13.223. Action for Agent. Help 'the action of helping' > 'any thing 
or person that affords help': Crist is eadmodegra help and ofermodigra 
fiell (893). — Aid 'action of aiding' > 'person aiding': The Lord that 
built the Earth and Skies is my perpetual Aid (1738). — Fare 'a going, 
journeying, etc' > 'a number of persons prepared for a journey, a troop, 
multitude': Brien bonnede his fare (= prepared his journey = prepared 
his companions and himself, 1205). — Failure 'the fact of failing to 
effect one's purpose' > 'a thing or person that proves unsuccessful': 
This attack was a failure also (1837). ^f V^'^ ^^^^^ ^^^ ^^^ ^ should turn 
out a failure (1865). Similarly success. The last two may be metaphor- 
ical, cf. II. 612 c. 

Latin senatus means not only the session, but also the assembly. 
Similarly consilium is sometimes used of a single person: Ovid. Trist. 
4. 2. 32: ille hortator pugnae consiliumque fuit. Statio originally means 
the 'standing', but also the person standing, the sentinel, as well as 
the place where he is posted (Stocklein 58 sqq.). 

For French instances, see Nyrop IV 190, 218, Carnoy 183. Cf. 
also Collin, Ata 77 sqq. for Swed. betjdning, bitrdde, regering, styng, 
and others. 

a change. That is a method not to be encouraged, except when absolutely neces- 
sary for lack of historical material. 

Leumann also gives some Latin instances. Fatum was originally 'that which 
is said or predicte'd'. He quotes equivocal instances from Cicero fat. 5: si Daphitae 
fatum fuit ex equo cadere "wenn es ihm geweissagt war", "wenn ihm ein Orakel- 
spruch war", "wenn es ihm Schicksal war", ib. 28 si fatum tibi est ex hoc morbo 
convalescere; ib. 30. si . . . ita fatum, erit: nascetur Oedipus Laio. 

372 GUSTAP STERN 13-224 

13.224. Action for Place of Action. Fare 'a going, journeying' 
;> 'a road, track, esp. the track of a hare or rabbit': Not a Hare Can 
he started from his fare (i6io). — Crossing is, secondarily, 'place of 
crossing': Statues... in the crossing of streets (1695). The ceiling 
is garnished, at the crossing and combining of the arches, with . . .(1828). — 

The type is very common, and also very old. Lat. accessus is shifted 
from 'the action of entering' to 'the place of entrance' in phrases like 
accessus est ad domum, accessu prohibere (Hey, ALL 9). Exitus and 
aditus have passed through a similar development, as well asF. entreee, 
issue, sortie, with equivocal instances as a la sortie de la messe, a l' entree 
du bain. (Cf. Nyrop IV 223, Carnoy 180 sqq., Collin, Ata 112, 114, 
Stocklein 58 sqq., Leumann, IF 45, iii — 112). 

13.23. Names of Qualities in Various Uses. The names of qualities 
are employed to denote persons or objects possessing these qualities. 
The shift is often metaphorical (cf. 11. 612 c) but apparently in some 
cases may also be explained as a permutation. (See for the whole 
problem, Hatzfeld, Objektivierung). 

The word quality is used, concretely and collectively, for 'people of 
good social position' (now archaic, vulgar or dialectal): Walk Bare- 
headed to his Master's Daughter, in imitation of Quality (1693). Another 
concrete meaning is 'title, description, character, capacity': Man agh 
to telle hir qualite, sib or freind or quat sco be (ai30o). 

Analogy with foreign patterns may be suspected in some cases, 
since the "collective" type is old and international. Cf. Latin juvenilis, 
F. jeunesse, Germ. Jugend, E. youth (See Carnoy 188). 

The use of such words to denote single individuals seems less com- 
mon. Youth 'young people' is Old English, but the sense 'young man' 
is later: Bi dat time dat he was sud, Wid faisered and strengthe kud (1250) . 

The names of qualities are also used to denote objects having that 
quality. Length is 'a long stretch or extent': Large lengths of seas and 
shores Betweene my father, and my mother lay (1595). Cf. Nyrop IV 
224 sqq. 

Consideration 'the taking into account of anything as a motive or 
reason' > 'something given in payment, a reward, remuneration, com- 
pensation': We gave them copper in consideration (sc. of services, etc. 

13.24. Names of Persons for Products, etc. We find in OE: seo 
hoc J>at is nemned Orosius. If the name of the author is written on the 


cover or title-page of the book, it is natural to use it, as we use the 
name labelled on any other object to denote it. I believe that this 
applies to a Milton, a Shakespeare 'a book or play by . . . ', and also 
to a picture, signed by the painter's name: a Rembrandt, a Romney. 
Equivocal instances are sometimes found: They have Moses and the 
Prophets (Bain I 190). In the same way wines are called by their la- 
bels: Beaune, Cliquot, J ohannisherger , etc. Instances like these are 
mostly classed as shortenings, and it is probable that shortening is the 
correct explanation in many cases, analogy or metaphor in others, 
but the possibility of a permutation should also be taken into consider- 
ation (cf. Wellander, Studien II 115, Stocklein 69, Carnoy 173 sqq., 
and 10.431 above). 

13.251. Place-name for Action or Event. Church 'a church building' 
> 'divine service': esp. in phrases like to attend church, go to church, be 
at {in) church, out of church, after church, etc. (see ISTED church sb. 11): 
Pu gast to chirche (1175). Similarly chapel, college, school, and others. 
This use is marked by the absence of the definite article. German 
Ki-rche shows a similar development (cf. Neumann IF 45, no, Stock- 
lein 60 qiioting also Schule, Tisch, and others. Efvergren 7 sqq.). 

The origin of the habit of using place-names for events is not clear. 
Wellander (Studien II 116) assumes shortening, and that explanation 
seems likely, at least in some cases. Cf. 11.332. I have no good equi- 
vocal instances, and permutation therefore should perhaps not be 
assumed as a possible explanation. 

Field 'open land, plain' acquires the special sense of 'battle-field', 
which > 'battle, victory'; cf. the phrases to fight, give, lose, make, win 
the {a) field, to get, have the field. The falde was hys & Arthourez (C1400). 
Of the fynd the maystry to haue. Of hym to wyn the fyld . . . Of hyme he 
wane the fyld j^at day (C1435. Cf. Efvergren 8). 

French greve has a complicated history. Out-of-work men used to 
assemble in the Place de Greve in Paris in order to offer themselves 
for employment. Faire Greve 'to walk about the Place de Greve' thus 
also meant 'to look for work'. This is a permutation. Through ade- 
quation, the latter element of meaning became predominant, to the 
exclusion of the association with the place, and greve came to mean 
'disemployment'. In the 19th century, the word came to be used of 
disemployment due to refusal to work, i. e., striking. In this special 
meaning, the word was dissociated from faire, and could be employed 

374 GUSTAF STERN 13-251 

with any suitable verb: se mettre en greve, declarer une greve, etc. (Cf. 
Nyrop, Das I,eben der Worter, quoted by K. O. Erdmann 24). The 
first permutation in this series of changes may be said to be conditioned 
by the ability of faire to take on two different meanings. In the first 
stage, faire means 'to walk about', in the second, 'to do a thing', cf. 
faire tin metier, faire le commerce. Possibly the association with the 
place shotild be assumed to disappear as the latter interpretation 
came to the front. 

Efvergren (42 sqq.) has an interesting explanation for the use of 
place-names to denote products of the place. He takes as a tj^pical 
instance Tars, originally a place-name, Tharsia, a country adjoining 
Cathay. The name also denoted a rich stuff, cloth of Tars. Of is used 
not only to indicate a place of origin, but also the material out of which 
something is made, as in howves of selk (Piers Plowm.). Consequently 
the locution tapytez of tars (Gawain 858), as it stands, could be inter- 
preted as showing either the origin or the material of the tapytez. The shift 
has been completed in dubbed in a dtiblet of a dere tars (Gawain 571) . 
Among the earliest place-names to be used of products in Knglish were 
the following, quoted by Efvergren 46: of tidy (sc. Toulouse) and tars 
(1360), wyn of Oseye (1362), wyn of gascoyne (1362), raysons of Coraunte 
(1390), a gown clothe of mostyrdewyk (1400). I do not think that the 
names of wines are likely to have arisen in this way, since of could scarce- 
ly be apprehended as denoting the material out of which the wine 
was manufactured, but with regard to other words the explanation 
appears plausible. 

13.253. Place-name for Inhabitants or Frequenters. We use the City 
for its inhabitants, the whole house for those that dwell in it, the gallery, 
the fit, the boxes, for those sitting there during a performance. It is an 
old type, and the explanation is often uncertain. It seems that metaphor 
is possible, but the presence of equivocal instances shows that per- 
mutations leading to the same result may also occur. If the develop- 
ment of the three last words is due to permutation, equivocal instances 
might have been of the type: applause from the pit, to play to the gallery, 
cf . NED gallery sb. 4 b: He addressed himself principally to his friends 
on his right and left, and in so inaudible a voice that his remarks did not 
reach the gallery (1817), which the NED interprets as signifying those 
sitting in the public gallery of the hall in question. 

For House 'legislative assembly', originally the building or room 


where they assemble, the JSTED has several eqtiivocal instances: The 
commons of the lower house, not forgettyng their olde grudge (1548). Sir 
Edward Coke is of the house (1624). To sit with the Peers in the higher 
House (1635). Cf. the Bar, the Bench, the Cabinet, the School: All f>e 
scole on him can wonder (1300); and see Efvergren 10 sqq. 

Nyrop (IV 198) quotes bureau and auditoire: "Pour les mots de ce genre 
les deux sens se presentent simultanement a I'esprit; c'est le contexte 
qui doit decider si par auditoire il faut entendre ou un local ou une 
reunion d'hommes". He further adduces un nid babillard, un parterre 
bienviellant, la ville s'est revoltee, toute la maison est en emoi, and con- 
clave, which is 'a room shut up with lock and key' > 'the room where the 
Cardinals shut themselves up to elect a Pope', > 'the assembly of the 
Cardinals for electing a Pope'. F. cour, from Latin cohortem, was first 
a country house, then a domain belonging to the king, then the resi- 
dence of the king and his attendants, and finally 'the persons attend- 
ing upon the king' (Cf. Carnoy 160, 260). 

Wellander (Studien II 116) quotes wdhrend dieser Zeit ordnete Peters- 
burg die allgemeine Mobilmachung an. Similarly we use Downing Street, 
Quay d'Orsay, Wilhelmstrasse for the governments or foreign depart- 
ments of the respective countries. I have explained the change of 
meaning of Moneta as a permutation (see 11. 14), and the same ex- 
planation seems applicable in many cases to instances like the above 
— always with regard to new formations; later instances may be in- 
fluenced by earlier patterns. If a man speaks of going to or coming 
from Downing Street, the meaning of the place-name is equivocal; the 
secondary meaning is at first determined by the context — the speak- 
er's interests or business, as known to the hearer — and later it may. 
become habitual, since it is the most important one in most connec- 
tions. A student will speak of his studies at Oxford, meaning the 
University, and the latter meaning of the word has become habitual 
among speakers whose interests in Oxford are with the University. 

13.36. Mental State for Object or Person causing it. These shifts 
are very common (see Hatzfeld, Objektivierung). It is not unlikely 
that some of them are to be explained as metaphors, but in many cases 
there are equivocal instances that point to permutation. Among them 
are the following. 

Care 'charge, oversight with a view to protection, preservation, or 
guidance' > 'an obi2ct or matter of care, concern, or solicitude': Gath- 

376 GUSTAF STERN 13-26 

ered the Princes . . . To taken counsell of their common cares (1590). The 
main care of any creature is self-preservation (1634). Cf. Bogholm 129. 

— Concern 'interest, solicitous regard, solicitude, anxiety' > 'a matter 
that concerns': 'Tis all mankind's concern that he should live (1700). 
Which are the common and greatest concern of all Christians (1732), — 
Delight 'the fact or condition of being delighted; pleasure, joy, or gra- 
tification felt in a high degree' > 'anything in which one takes delight': 
It es a place of delytez (1400). Why, Sir lohn, do you thinke . . . that 
euer the deuill could haue made you our delight? (1598). — Desire 'the 
fact or condition of desiring' > 'an object of desire': Hee hoped to haue 
there of his hertes desyres (1340 — 70). He sawe that he ne myght nought 
acheuen hys desyre (1413). — Hate > 'the object of hatred': poetical^ 
according to the NED: My onely Loue sprung from my onely hate (1592). 

— Pride > 'that of which any one is proud': My seyntuarie, the pryde 
of sour empyre, and desyrable thing of sour eyen (1382). Her f>e pryd 
of water ford felle, her all hys myght went to noght (1425). — Scandal > 
'a person whose conduct is a gross disgrace to his class, country, posi- 
tion, or the like': Duncombe, a drunken M. A. of St. Marie Hall, a scan- 
dall to his profession (1683). — Trouble > 'a thing or person that gives 
trouble': Alack, what trouble was I then to you? (1610). Your appointed 
Feasts . . . are a trouble vnto me, and I am weary to beare them (1611). 

French and Latin influence is possible in some of these cases. 

13.3. Adjectives and Adverbs. 

Adjectives signifying 'rapid', in the physical sense, acqmre the 
meaning 'eager, willing' when used of living beings or other objects 
conceived as living. The apprehension shifts from the purely physical 
notion to the notion of the mental state that is considered as causing, 
or as involved in, the physical rapidity. This happens with OE. hrced, 
swift, snel and leoht. The following are equivocal instances, all of them 
taken from Stern, Swift (261, and the paragraphs quoted there): Hrede 
(Vulgata: veloces) foet heara to ageotenne blod. Ic Icere, pcet fiu beo hrcedra 
mid hreowlicum tearum. We beod ful swyfte to farenne geond ealle wid- 
gylnyssa Codes rices. Hirce fet hrcB^e I snelle to cBgiotcencB I to scedende blod. 
Se wees mid dcsdum snelra J)onne he mcegenes hcefde {: celeritate magis 
quam virtute fretus). Se hcefde moncynnes mine gefrcege leohteste hond 
lofes to wyrcenne. Patt ajj wass lihht all allse chaff To folljhenn alle 
sinnes (1200). 


The use of adjectives with nouns that denote living beings is the 
cause of several other sense-changes in adjectives, as 'rapid' to 'rash'; 
'rapid' or 'prompt' to 'mentally quick'; and 'ready, willing' to 'prompt'. 
See, for particulars, Stern, Swift 261 sqq. No doubt the adjectives 
are often thus influenced by the nature of the governing word, but there 
is not sufficient material available to show if this is a general rule. 

German billig was originally 'sequus', and billiger Preis meant 'a 
reasonable, moderate' price, a price that was not exaggerated, but 
filled the demands of csquitas. Since such a price is generally com- 
paratively low, the word billig was repeatedly associated with the notion 
of 'low', which thus became an element of meaning, and finally took 
the place of predominant element, as being frequently the most im- 
portant element, both to seller and buyer. When the seller said, ich 
verkaufe es dir um einen billigen Preis, his billig might be only relatively 
'moderate', but for the buyer it was not 'moderate' unless it was 'cheap', 
and if they were to come to an understanding the seller would have 
to accommodate himself to the buyer's notions (Stocklein 19. Cf. 
Wellander, Studien III 244, and 10.424 above). 

The permutations of adverbs are often dependent on the meanings 
of the governing verbs, like those of adjectives on the nouns. ME 
faste originally means 'vigorously, energetically'. It was employed 
to intensify verbs denoting some kind of physical action, and was one 
of the most common intensifiers in MK. In connection with verbs 
of motion, as he renneth faste 'he runs vigorously, with energy', the 
adverb took on the meaning 'swiftly', since anyone who runs with all 
his might will also get over the ground swiftly. That is often the most 
important point for both speaker and hearer, and so the new meaning 
was enabled to take the place of predominant element, to the exclusion 
of the earlier meanings. (Cf. Stern, Swift 265 sqq.). Several other 
adverbs pass through a similar development. 

Another change of the present type is that from 'rapidly' to 'imme- 
diately' which has been described in 7.45. I may point out that 
here, too, the change in the adverb is correlated to a change in the 
meaning of the governing verb, from imperfective to perfective. 

13.4. Verbs. 

To cancel, from F. canceller, from L. cancellare, is "to make lattice- 
wise, to cross out a writing, f. I^. cancellus, cancelli, crossbars, lattice. 


In English the original meaning was 'to deface or obliterate (writing) 
by drawing lines across it lattice- wise, to cross out, strike out. Of 
legal documents, deeds, etc.: To annul, render void or invalid by so 
marking". But the effect of the action was, as often, the most im- 
portant element in the meaning of the word, and came to predominate, 
to the exclusion of the original concrete meaning: CanceUynge or strek- 
ynge owte a false word (1440). This may be equivocal. The new 
meaning (after adequation) is shown clearly in: All such handes and 
promysses that the Kynge or any other had made . . . shuld he adnulled 
& cancelled (1494). 

To soothe is related to the adjective sooth 'true', and came to mean 
'to declare to be true' > 'to back up or support a person in a state- 
ment or assertion'. Since this method is often resorted to in order 
to please or humour somebody, the verb took on the sense 'to blandish, 
cajole, or please a person by agreement or assent, to flatter in this way, 
to humour': For sooth Roister Doister in that he doth say, And require 
what ye will ye shall haue no nay (Ralph Royster Doyster, Arb. I i: 
'for back up (or humour) R. D. in what he says, and you shall not 
be refused whatever you ask'). 

The earlier meaning of want is 'to be in want of something implied 
by the context'; this passed into 'to suffer the want of, to have occasion 
for, need, require, to stand in need of: Deyr cusyng, pray I the, Quhen 
thow wantts gud, com fech ynewch fra me (1470). From the modern 
point of view this would naturally be taken to mean 'when you desire', 
but that sense is not given by the NED until 1706: All such as want to 
ride in Post-haste from one World to the other. The equivocation was 
there from the beginning, for what a person needs, if it is something 
necessary, he also desires. 

13.5. Particles. 

There are in English some participles in -ing which develop a con- 
junctional or quasi-conjunctional meaning, e. g., seeing 'considering 
the fact that, inasmuch as, since, because'. / wol . . . exhorte you to 
take it as . . . paciently as ye can, seeyng that we al be mortal and borne 
to dey (1503). The NED remarks that this "is a doubtful or transitional 
example, as the pple. admits of being construed as in concord with 
the subject, in the sense 'recognizing, perceiving'. The development 
of^ the conjunctional use may have been aided by the similarity of sound 


with sen, sin, conj." Other participles with a similar development 
are considering, excepting, providing, supposing (of. ISTED and B0g- 
holm 141). It is common to them all that the notion of the verbal ac- 
tion disappears, and the notion of the logical relation between the two 
relevant clauses or words takes its place as predominant. I am not 
quite certain that this cannot be an adequation. 

Carnoy (236) gives the following instances in illustration of the 
development of F. cependant from an adverb of time to an adversative 
conjunction: // me recommande d'agir ainsi, cependant [sc. qu'il parte) 
je I'ecoute poliment, II me recommande d'agir ainsi, cependant (equi- 
vocal) je prete pen d' attention a ses propos. II me recommande d'agir 
ainsi, cependant je n'en ferai rien. 

The development of il a beau is thus illustrated: // a beau rire ( = 
it pent rire a I'aise), je ne puis I'attendre. II a beau rire, cela ne 
m^'emeut guere. II a beau rire, je ne I'en attraperai pas moins. The 
last phrase shows the antithetic meaning (Carnoy 236). 

German instances of a similar development are weil and wdhrend: 
Die Blumen schliessen sich, weil es regnet; here the causal notion 
may enter into the meaning of the originally temporal conjunction. 
Similarly, wdhrend die Belagerten schwere Not litten, hatten die Belagerer 
Ueberfluss, where an adversative sense is suggested, and may develop 
into the predominant element. (See Neumann 109). 

In English, while shows a similar development from temporal to 
adversative and concessive import: Painefully to poare vpon a Booke, 
To seeke the light of truth, while truth the while Doth falsely blinde the 
eye-sight of his looke (1588). While they deny a Deity, they assert other 
things on far less reason (1662). While here means 'when on the contrary 
or on the other hand, whereas'. (I^atin instances are given by Stock - 
lein 15 sqq.) 

Another group is constituted by local prepositions developing a 
temporal meaning. Stocklein (56) quotes Latin statim, extemplo, illico 
(in loco), German a